Consistent client-side cache

CROSS-REFERENCE TO RELATED APPLICATIONS

0001The present application claims the benefit of U.S. Provisional Application No. 60/853,117, filed on Oct. 20, 2006.

FIELD

0002Embodiments of the invention relate to computer systems, and more particularly to a client-side cache.

BACKGROUND OF THE INVENTION

0003A cache is a collection of data that is a duplication of original values stored elsewhere or computed earlier, when the original data is expensive to fetch or to compute relative to retrieval from the cache. For example, a server-side query cache for a database may store the results for a Structured Query Language (SQL) query received by the server in a cache in server memory. Storage of the query results in the server-side query cache enables the retrieval of the results for a query upon receipt of an identical query without requiring the server to re-execute the query against the database. In order to receive accurate results from the query cache, there must not have been a change in the data that was relied upon in the formation of the initial result for the query. A server side query cache may improve the retrieval results of the data, primarily with read only or read mostly data.

0004In one embodiment, the cache is implemented as a daemon process, a process running in the background, running on the client that the client interacts with for caching. In another embodiment, a caching daemon process may run on a middle tier and be shared between multiple clients. The cache can be located per client session, per client process shared by different sessions in same process, in shared memory/daemon-process on client shared by different client processes on the same machine, or in a daemon on a different machine shared by different client machines. A client-side query cache can either be in memory and/or on physical storage accessible by client processes.

0005Client-side query caches, a query cache in client memory, provide an additional benefits over a server-side cache. First, caching on the client eliminates the necessity to perform the request to the server and receive the response from the server in order to retrieve the query results thereby improving response time. Client machines can keep being added horizontally to provide the caching capabilities in client memory and reducing the expense of setting up additional servers to support caching query result. Further, storage on the client side offers the benefit of not only having the queries closer to the client but also ensures that the most relevant queries to the client are stored at the client.

0006However, storage of the query results in a client-side query cache may introduce data consistency problems that are not present with the use of a server-side query cache, and the data correctness problems, if left unresolved, produce unexpected results for the user querying the database. In the database, a snapshot, a record of the state of the database, is created when a transaction is executed that changes the state of the database. The snapshot is monotonically increasing and there is never a regression back to an earlier snapshot, which means that succeeding queries in time see more recent snapshots and never earlier snapshots. Any statement executed on the database is guaranteed to run against such a consistent snapshot, also known as the execution snapshot, that guarantees to include all changes to the state of the database done by all transactions leading up to the snapshot, and no changes to the database after the snapshot will effect the results of the query run against that snapshot. The database guarantees the results of the query are generated against the snapshot of the database at the time of receipt of a query (also known as the query execution snapshot), and the user expects query results from a cache to maintain this level of transactional consistency. At the server, the server-side query cache can simultaneously invalidate query results in the cache upon receipt of a transaction that necessitates invalidating the corresponding query results stored in the cache. The client-side query cache residing on the client is not able to simultaneously invalidate the cache with changes that occur in the database, hence the challenge lies in the ability to produce consistent query results with the use of a client-side cache.

0007FIGS. <b>1</b>A-B are block diagrams that illustrate the consistency problems encountered in the described approach with client-side query caches. In <figref idref="DRAWINGS">FIG. 1A</figref>, SQL queries have been previously requested by the Client <b>100</b> with the Database Application Programming Interface (API) <b>102</b> and the Client-side cache <b>104</b> has stored query results, as depicted with Query Results for Table A <b>106</b> and Query Results for Table B <b>108</b>. The Query Results for Table A <b>106</b> and Query Results for Table B <b>108</b> in the Client-side Cache <b>104</b> reflect the contents of the tables, Table A <b>110</b> and Table B <b>112</b> respectively, currently in the Database <b>114</b> on the Database Server <b>116</b>. There is a relationship between Table A <b>110</b> and Table B <b>112</b> (e.g. trigger) that requires that a portion of the data in Table A <b>110</b> be placed in Table B <b>112</b>. With the Database API <b>102</b>, in the same transaction that modifies Table A <b>110</b>, Client <b>100</b> makes a Request to Insert Mehul in to Table A <b>118</b>.

0008<figref idref="DRAWINGS">FIG. 1B</figref> shows the contents of the Client-side Cache <b>104</b> of the Client <b>100</b> and the Database <b>114</b> after the Database Server <b>116</b> responded to the Request to Insert Mehul To Table A <b>118</b>. The Database Server <b>116</b> has responded to the Request to Insert Mehul <b>118</b> to Table A <b>118</b> by inserting a second row to Table A <b>110</b> (e.g. “2 Mehul mehulB”) and the insertion of the row to Table A <b>110</b> has triggered the addition of a second row to Table B <b>112</b> (e.g. “2 Mehul”). The Client-side Cache <b>104</b> is aware of the request made by the Client <b>100</b> as reflected by the contents of the Query Results for Table A <b>106</b> and unaware of the addition to Table B <b>112</b> as reflected by the Query Results for Table B <b>108</b>. Thus, in Figure B, if the Client <b>100</b> in <figref idref="DRAWINGS">FIG. 1B</figref> requests the contents of Table B <b>112</b>, then the Client <b>100</b> will lookup the results in the Client-Side Cache <b>104</b> and retrieve the Query Results for Table B <b>108</b> without the newly added row. The production of query results for Table B with the client-side cache without the newly added row cannot be properly handled by the application relying on the data.

0009Thus, there is a need for a solution to ensure the same level of consistency with client-side cache as the user expects with the database or the use of a server-side cache. The solution should be both a transparent solution and guarantee transactional correctness similar to that provided by the database with the use of a snapshot. As another example, query results could be a join of multiple tables and there is a need to refresh cached result(s) with database changes that affect any of the tables in the query. Additionally, there could be different clients or software running on the server that concurrently make database changes that affect the cached result set and there is a need to identify all database changes that affect cached result sets on the client. Beyond database changes, user environment settings (e.g. changing the language from French to German) may affect the result set and there is a need to detect such non-database changes to refrain from returning incorrect results to the application. A change in session or environment settings may indicate a need to invalidate cached result sets or create new cached result sets.

0010Although embodiments are described in reference to a client-side query cache, it should be noted that the consistent caching implementation can also be used with caches that support other content. For example, the consistency of the client-side cache can be used to ensure consistent caching of any other type of cached content that may be derived from the result from of a database operation.

SUMMARY OF THE INVENTION

0011A method, system, and computer program product for caching results in a client-side cache is described. In some embodiments, the caching is performed by associating a first snapshot of a database with a client, wherein the first snapshot indicates a state of the database after a last database request by the client, and indicating any number of cached results as being invalid in the client-side cache for the client based upon the first snapshot.

0012A computer program product with a computer usable medium having executable code to execute a process for implementing a client cache in a computing system is described. In some embodiments, a client cache process is performed by associating a first snapshot of a database with a client, wherein the first snapshot indicates a state of the database after an interaction with the database by the client, and invalidating any number of invalid cached results in the client cache for the client based upon the first snapshot.

BRIEF DESCRIPTION OF THE DRAWINGS

0013The invention is illustrated by way of example and not limitation in the figures of the accompanying drawings, in which like references indicate similar elements and in which:

0014FIGS. <b>1</b>A-B show block diagrams of the prior art.

0015<figref idref="DRAWINGS">FIG. 2A</figref> is a flowchart of a process for implementing a consistent client-side cache.

0016<figref idref="DRAWINGS">FIG. 2B</figref> depicts an architecture for implementing consistent query caches.

0017<figref idref="DRAWINGS">FIG. 3</figref> is a flowchart of a process for implementing a consistent client-side cache.

0018<figref idref="DRAWINGS">FIG. 4</figref> is a block diagram for an implementation of a consistent client-side cache.

0019<figref idref="DRAWINGS">FIG. 5</figref> is a flowchart of a process for implementing a consistent client-side cache.

0020<figref idref="DRAWINGS">FIG. 6</figref> is a block diagram for an implementation of a consistent client-side cache.

0021<figref idref="DRAWINGS">FIG. 7</figref> is a flowchart of a process for an implementation of a consistent client-side cache.

0022<figref idref="DRAWINGS">FIG. 8</figref> is a flowchart of a process for an implementation of a consistent client-side cache.

0023<figref idref="DRAWINGS">FIG. 9</figref> is a flowchart of a process for an implementation of a consistent client-side cache.

0024<figref idref="DRAWINGS">FIG. 10</figref> is an architecture for support of an implementation of a consistent client-side cache.

DETAILED DESCRIPTION

0025Embodiments of the present invention provide methods, systems and mediums for implementing a consistent client-side cache.

0026<figref idref="DRAWINGS">FIG. 2A</figref> is a flowchart for a process for implementing a client-side cache. In <figref idref="DRAWINGS">FIG. 2A</figref>, a snapshot of a database is associated with a Client (<b>201</b>) that is supported by a client-side cache. In one or more embodiments, the snapshot may be stored at a cache manager for the client-side cache, on a database server, in client memory, or in any storage that is accessible by a module that invalidates cached results provided that there are cached results to invalidate. Based upon the association of the client with the snapshot, the invalid cached results in the client-side cache may be indicated based upon the snapshot (<b>203</b>). Embodiments of the client-side cache may compare the snapshot associated with the client with the current snapshot of the database to determine the cached results that can be invalidated. In one or more embodiments, a database server or database may store all transactions that occur on the database and reference the stored transactions to determine the changes to the database over time after the snapshot.

0027<figref idref="DRAWINGS">FIG. 2B</figref> shows an example of a client server architecture for implementing a consistent client-side query cache. In <figref idref="DRAWINGS">FIG. 2B</figref>, Client <b>200</b> may support any number of Sessions <b>1</b> . . . N <b>202</b> that use the Database API <b>204</b> to make requests to the Database Server <b>206</b>. Sessions <b>1</b> . . . N <b>202</b> are lasting connections with a server. The Database Server <b>206</b> may access or consist of the Database <b>208</b>, a collection of data, to fulfill SQL query requests or DML commands from the Client <b>200</b>. In one or more embodiments, the Database Server <b>206</b> could access or consist of a cluster of databases and within the cluster broadcast received transactions to the other database instances within the cluster. In response to SQL query requests, the Database Server <b>206</b> will return the result set generated upon execution of the SQL query by the Database <b>208</b> and may determine that the query is cacheworthy, a good candidate for caching.

0028In general for some embodiments, a cacheworthy query is a query that will be repeated and the data returned as a result of the query runs less of a risk of becoming stale (e.g. a query that requests the current date and time value is not cacheworthy because the date and time will be stale upon retrieval from the cache). Other considerations in determining the cacheworthiness of the query may include: the size of query results (e.g. results may not fit in the client cache), whether the query results rely on data from read only or read mostly tables, and the frequency expectation for a repeat of the query request. Alternatively, considerations for designating a query or a table on which queries are not to be cached may include: if the table is expected to change often, if the query is very inexpensive to compute, if the query is not repeated often or if the result set for the query is very large. Those skilled in the art will recognize that there are many factors or considerations that may contribute to whether a query or a query on a particular table should not be cached.

0029Queries and queries on a particular table that are cacheworthy or should not be cached can be determined both in design of the database and at runtime of the consistent client-side cache. In one or more embodiments of the present invention, a database query optimizer can be used to transparently detect cacheworthy query candidates and make a client aware that a query is a good candidate for caching. Embodiments may use manual hints in Data Description Language (DDL) commands that define tables or views to indicate queries that are cacheworthy or not cacheworthy. A hint may also be provided in the SQL text of a query. For example, a hint may be provided for a table T to indicate that queries on table T are cacheworthy when table T is a read only or read mostly table. A hint that designates a query or queries on a particular table does not necessarily imply that the query will be cached in all cases. Embodiments may also indicate programmatically with the use of the database API that a query is cacheworthy. Those skilled in the art will recognize that there are many factors that contribute to whether a query will be cached such as the actual size of the result set and the amount of memory available in the cache. In one or more embodiments, tables or queries can be designated as not to be cached with the use of DDL commands, SQL queries or programmatically.

0030Continuing with <figref idref="DRAWINGS">FIG. 2B</figref>, if the query is not cacheworthy, then the query results are returned to the Database API <b>204</b> without registration with the Database Change Notification Module <b>210</b>. Alternatively, if the query is cacheworthy, then the Database Server <b>206</b> will register the query with the Database Change Notification Module <b>210</b>. Examples of approaches for a Database Change Notification Module are described in co-pending U.S. application Ser. No. 10/866,344, filed on Jun. 10, 2004, entitled “Query-Based Invalidation Subscription”, which is hereby incorporated by reference in its entirety and co-pending U.S. application Ser. No. 10/866,433, filed on Jun. 10, 2004, entitled “Active Queries Filter Extraction”, which is hereby incorporated by reference in its entirety.

0031The Database Change Notification Module <b>210</b> provides notification of changes in the underlying data relied upon to generate the query results for registered queries that would cause the query results stored in the Client-side Cache <b>212</b> to be invalid. The Database Change Notification Module <b>210</b> generates a set of cache invalidations for registered queries that are pertinent to the Client <b>200</b> and/or the Session <b>202</b> used for the current server roundtrip (i.e. make requests and receive responses from a database server), and the cache invalidations will be returned whenever the Client <b>200</b> sends any message to the Database Server <b>206</b>. In one or embodiments, the invalidations are put in the shared memory of the instance allowing access to all processes running on the instance to the invalidations and to other shared memory instances before the transaction commits.

0032In some embodiments, a cached result identifier in the set of cache invalidations is the combination of a query id and a transaction id. To differentiate result sets, each result set may be assigned a unique identifier by the Database Server <b>206</b>, referred to as a cached result identifier or a query Id. Multiple cached result sets in a Client-side Cache <b>212</b> may be associated with the same query Id. In one or more embodiments, a query Id may be combined with a sequence number that is incremented for every Client-side Cache <b>212</b>. In one or more embodiments, the last query Id may be stored persistently so that the sequence number is available after a Database Server <b>206</b> restart. Optionally, depending on the granularity of the Database Change Notification Module <b>210</b> described below, the query Ids can be shared across SQL-text, across bind values and across Client-side Caches <b>212</b>. In one or more embodiments, the query Id can be shared across Client-side Caches <b>212</b> to efficiently utilize database resources and allow cleanup of the Query Ids stored at the Database to be based on a reference count of the number of Client-side Caches <b>212</b> receiving change notification for the SQL queries with the same Query Id.

0033Continuing with <figref idref="DRAWINGS">FIG. 2B</figref>, the Database Server <b>206</b> will also return an Out Snapshot, a snapshot of the database after receipt of a database request from the Client <b>200</b>, with the query results, and the cache invalidations to the Database API <b>204</b> up to the Out Snapshot may be included in the returned message.

0034The Cache Manager <b>214</b> invalidates the identified cached result sets corresponding to the query ids included in the cache invalidations for the Client <b>200</b> and the Session <b>202</b>. In some embodiments, a query id corresponds to an identifier for the query assigned by the Database <b>208</b>. Upon receipt of the cache invalidations, and the Out Snapshot from the Database Server <b>206</b>, the Database API <b>204</b> sets the Visible Snapshot <b>216</b>, a record of the state of the database at the time of the last database request (e.g. query, DML request) for the Client <b>200</b>, associated with the Client <b>200</b> to the Out Snapshot, and passes the cache invalidations to the Cache Manager <b>214</b>. The Database API <b>204</b> passes the returned query results to the Client <b>200</b>. The Database API <b>204</b> passes the query results to the Cache Manager <b>214</b> provided that the query results are cacheworthy and the Cache Manager <b>214</b> caches the query results in the Client-side Cache <b>212</b>. If the query is cacheworthy, then the query results will contain the keys (e.g. a compile time key and a run time key described in more detail below) for entry into the cache, and retrieval of the query results from the cache.

0035In one or more embodiments, a query may be deemed cacheworthy by the Database Server <b>206</b> but may not be cached by the Client <b>200</b> due to additional constraints on caching for the particular Client <b>200</b> (e.g. resource constraints limiting the size of result sets). The constraints (e.g. maximum cached result set size) may be specified at the Database Server <b>206</b> and communicated to Client-side Cache <b>212</b>. These constraints may also be specified at client. To prevent the Database Server <b>206</b> from continuing to send invalidations for the query ids related to the uncached results, the Client <b>200</b> may indicate to the Database Server <b>206</b> on a subsequent request a list of query ids that are not to be cached on the Client <b>200</b>.

0036A database query optimizer may find that the query is no longer cacheworthy. For example, the database query optimizer may determine that the result set is changing too often and requiring too frequent updates to the results. In such cases, the client-side cached result set will be invalidated and may not be subsequently cached.

0037At a later point in time, the Database API <b>204</b> may call the Cache Manager <b>214</b> to lookup the query results in the Client-side Cache <b>212</b> upon receipt of a query request, and will return the query results from the Client-side Cache <b>212</b>. The Cache Manager <b>214</b> will incorporate user environment settings such as the character set encoding and language into the lookup so the application receives the correct result sets in the same way that the client would receive the result sets without the client cache enabled. Although <figref idref="DRAWINGS">FIG. 2B</figref> illustrates a client server architecture, those skilled in the art will recognize that embodiments are not limited to a client server model. In one or more embodiments, a client-side cache may be implemented on an n-tier architecture or with use of any remote networked node that caches data.

0038In one or more embodiments, the Client <b>200</b> may not be making periodic requests to the Database Server <b>206</b> such that it is possible for lookups to the Client-side Cache <b>212</b> to return consistent data with respect to the Client <b>200</b> Visible Snapshot <b>216</b>, but the results may be stale in comparison to the changes on the Database <b>208</b>. For this embodiment, the Client <b>200</b> Visible Snapshot <b>216</b> may lag behind the Database <b>208</b> current snapshot to an extent that is unacceptable to the Client <b>200</b> application. One approach to handling a possible lag on the Client <b>200</b> is to set a maximum Client-side cache <b>212</b> lag parameter on the Database Server <b>206</b> that may be communicated to the Client-side cache <b>212</b>, and/or set a cache lag parameter on the Client <b>200</b> itself. A Database Server <b>206</b> response received by a Client <b>200</b> may update a Cache-Server interaction time. If the difference between a current time and the Cache-Server interaction time is greater than a maximum client-side cache lag parameter, then the Cache Manager <b>214</b> may not perform a lookup of the Client-side Cache <b>212</b> and may instead force the query to be executed against the Database Server <b>206</b>. The response from the Database Server <b>206</b> will update the Cache-Server interaction time.

0039<figref idref="DRAWINGS">FIG. 3</figref> is a flowchart for a process for implementing a consistent client-side cache. <figref idref="DRAWINGS">FIG. 3</figref> describes the implementation in regards to a database request that may be implemented with the use of any type of request made to the database. The caching implementation may be applied for use with any database request originating from client cache process server, including but not limited to database calls related to large objects (LOBS), complex objects, XML calls, and checking the server status. Initially, a Client <b>200</b> makes a database request with the Database API <b>204</b> (<b>300</b>). Next, if the Client <b>200</b> request does not involve a database query (<b>301</b>), then the Database API <b>204</b> makes a Database Server <b>206</b> request (<b>303</b>). Next, the cache is updated with the Cache Invalidations (if any) returned by the Database Server <b>206</b> (<b>320</b>). To update the cache, the cached queries included in the set of Invalid Query Ids are removed from the cache. In one or more embodiments, the Database API <b>204</b> will return with correct query results for the Invalid Query Ids. The identified set of Dirty Query Ids are used to indicate that the session should not use the cached copy of the data due to the uncommitted changes made to the database tables by the session. Next, the Database Server <b>206</b> returns the results to the Client <b>200</b> (<b>310</b>).

0040Alternatively, if the database request is a database query (<b>301</b>), then the Database API <b>204</b> requests that the Cache Manager <b>214</b> lookup the query results in the Client-side Cache <b>212</b> (<b>302</b>) which is shown in <figref idref="DRAWINGS">FIGS. 4 and 5</figref>.

0041<figref idref="DRAWINGS">FIG. 4</figref> is a block diagram for an interaction between a Cache Manager and a Client-side Cache. To perform the cache lookup (<b>400</b>), the Cache Manager <b>214</b> passes parameters that are combined to form a key to lookup the query in the Client-side Cache <b>212</b>. In an embodiment of the invention, a compile time key, created with the SQL text and user information that does not change when executing a query repeatedly (e.g. preferred language, user identification), is combined with a run time key, formed with the SQL query bind values that may change between queries (e.g. Select*from Table A where name=[bind value];), to lookup the query results in the cache. The runtime key may also include user environment settings such as character set encoding to receive character values, language settings, and time zone. If the cache lookup results in a cache hit (<b>402</b>), then the Cache Lookup (<b>400</b>) returns the result set.

0042<figref idref="DRAWINGS">FIG. 5</figref> is a flowchart for a process for a Database API request to the Cache Manager to lookup the query results. Initially, the compile time key is created from the SQL query text and the user information (<b>500</b>). Embodiments of the compile time key may be created at the client during query execution or the server may create the compile time key and send it to the client during query execution. A run time key formed from the bind values and environment settings may be created (<b>502</b>) and the Cache Manager <b>214</b> combines the compile time key and run time key to do the lookup (<b>504</b>) in the client-side cache <b>212</b>. At <b>506</b>, the Cache Manager determines if the query results are in the cache. If the query results are not in the cache, then the Cache Manager <b>214</b> returns a Cache Miss (<b>508</b>). The first lookup for a particular query will result in a cache miss. Alternatively, if the Cache Manager <b>214</b> retrieves the query results, then the Cache Manager <b>214</b> returns the cached query results and returns a cache hit (<b>510</b>).

0043Continuing with <figref idref="DRAWINGS">FIG. 3</figref>, if there is a cache hit (<b>304</b>), then the query results are in the Client-side Cache <b>212</b>, following the Cache Manager <b>214</b> lookup (<b>302</b>), then the Cache Manager <b>214</b> retrieves the query results from the Client-side Cache <b>212</b> (<b>306</b>) and sets the Query Results to the results returned from the Client-side Cache <b>212</b> (<b>308</b>). The Database API <b>204</b> returns the results to the Client <b>200</b> (<b>310</b>).

0044Alternatively, in <figref idref="DRAWINGS">FIG. 3</figref>, if there is no cache hit (i.e. a cache miss) at (<b>304</b>), then the Database API <b>204</b> requests query results from the Database Server <b>206</b> (<b>312</b>) described in more detail below in <figref idref="DRAWINGS">FIGS. 6 and 7</figref>. At (<b>314</b>), the Query Results are set to the results returned from the Database Server <b>206</b> and a determination is made as to whether the query is cacheworthy (<b>316</b>) as described above with <figref idref="DRAWINGS">FIG. 2</figref>. If the query is cacheworthy, then the Cache Manager <b>214</b> caches the Query Results (<b>318</b>) and the Cache Manager <b>214</b> is updated with the Cache Invalidations (if any) returned from the Database Server <b>206</b> (<b>320</b>). In one or more embodiments, the Cache Manager <b>214</b> caches the query id, result set, and bind values for the cacheworthy query. Additionally, embodiments of the invention may cache the Out Snapshot of the query associated with the cached results. Alternatively, if the query is not cacheworthy (<b>316</b>), then the cache is updated with the Cache Invalidations (if any) returned by the Database Server <b>206</b> (<b>320</b>). To update the cache, the cached queries associated with the set of Invalid Query Ids are removed from the cache. In one or more embodiments, the Database API <b>204</b> will return with correct query results for the Invalid Query Ids. The identified set of Dirty Query Ids and their associated cached results are used to indicate that the session should not use the cached copy of the data due to the uncommitted changes made to the database tables by the session. Next, the Database Server <b>206</b> returns the results to the Client <b>200</b> (<b>310</b>).

0045<figref idref="DRAWINGS">FIG. 6</figref> is a block diagram for an interaction between a Database API request for results from a Database Server. The Client <b>200</b> will access the Database <b>208</b> with any Database API <b>204</b> used by those skilled in the art (e.g. Oracle Call Interface (OCI), Java Database Connectivity (JDBC)). In an embodiment, the Database API <b>204</b> used to implement the consistent client-side query cache is an extension of a remote procedure call within a database API. Database API <b>204</b> makes any request <b>600</b> by passing parameters of an In Snapshot and the request. In an embodiment, the In Snapshot parameter is set to the client's Visible Snapshot <b>216</b>, a snapshot indicating the last interaction with the database, and the Query parameter is set to the client's desired SQL query. The Database Server <b>206</b> responds by returning the result for the request from executing the database request against the database, the Out Snapshot, and the Cache Invalidations, including the Invalid Query IDs, and the Dirty Query IDs (<b>602</b>). The Out Snapshot is a latest snapshot of the database. The Cache Invalidation is described in more detail in the description of <figref idref="DRAWINGS">FIG. 7</figref>.

0046<figref idref="DRAWINGS">FIG. 7</figref> is a flowchart of the process of a Database API request to a Database Server. Initially, the Database API <b>204</b> parameter In Snapshot is set to the Client <b>200</b> Visible Snapshot <b>216</b> (<b>700</b>) to ensure that the Database Server <b>206</b> is aware of the state of the Database <b>208</b> at the last interaction that the Client <b>200</b> had with the Database <b>208</b>. Next, the Database Server <b>206</b> determines whether the database request involves a database query (<b>701</b>). If the database request is not a query (<b>701</b>), then the Database Server <b>206</b> requests Database Change Notification generate Cache Invalidations and Out Snapshot as described in more detail with <figref idref="DRAWINGS">FIG. 9</figref>. If the database request is a query, then the Database Server <b>206</b> determines whether the query is cacheworthy (<b>702</b>) as described above in regards to <figref idref="DRAWINGS">FIG. 2</figref>. If the Database Server <b>206</b> determines that the query is not cacheworthy (<b>702</b>), then the Database Server <b>206</b> requests that the Database Change Notification Module <b>210</b> only generate the Cache Invalidations (if any) and Out Snapshot (<b>706</b>) described in more detail with the description of <figref idref="DRAWINGS">FIG. 9</figref>. Alternatively, if the Database Server <b>206</b> determines that the query is cacheworthy (<b>702</b>), then the Database Server <b>206</b> registers the query with the Database Change Notification Module <b>210</b> (<b>704</b>) described in more detail with the description of <figref idref="DRAWINGS">FIG. 8</figref>.

0047<figref idref="DRAWINGS">FIG. 8</figref> is a flowchart of a process for Database Change Notification Module Registration. By registering the query with the Database Change Notification Module <b>210</b>, the Client <b>200</b> is able to receive notification of any changes that occur that would affect the query results stored in the Client-side Cache <b>212</b>. Registration begins by assigning the value of the query id in the Database Change Notification Module (<b>802</b>). In one or more embodiments, the query id allows for identification of both the query and the client that requested the query. In one or more embodiments, a database clustering approach may be implemented and the Database Change Notification Modules registration information must be broadcast to the database instances within the cluster.

0048Registration may include determining the granularity specification for invalidation of the client-side query cache contents (<b>804</b>). For example, the client can specify that the invalidations of the query results be done when there is a change to any table referred to in query (i.e. coarse granularity) or the invalidation could be done to the query results only if there is a change in the result set only (i.e. finer granularity). For coarse grained notifications, registration involves providing the SQL text for the query and other information such as the user executing the query to the Database Change Notification Module <b>210</b>. Since the process of registration involves a certain amount of overhead, repeated registrations in coarse grained notification can be eliminated. For example, in one or more embodiments, the Database Change Notification Module <b>210</b> may recognize that the SQL text has already been registered by a given Client <b>200</b> for a user. In the case of fine grained notification, the query has to be registered every time because notifications are desired only when the result set is invalidated and not just when an underlying table changes.

0049Those skilled in the art will recognize that the granularity specified at registration could be at variety of levels. Examples of available granularity options are: database level (i.e. detect all changes to the database), schema level (i.e. detect changes to the schema for the user), object level (i.e. detect changes to tables), projection level (i.e. detect changes to columns), selection level (i.e. detect changes to rows), and result set level (i.e. detect changes to a query result set). Depending on the granularity specification for invalidation of the client-side query cache contents at the Database Change Notification Module <b>210</b>, the query Ids in some embodiments can be shared across SQL-text, across bind values and across client caches. In one or more embodiment, the query Ids can be shared across client caches to efficiently utilize database resources.

0050The cleanup or garbage collection of the query Ids stored on the Database may be based on a reference count of the number of Client-side Caches <b>212</b> receiving change notification for the SQL queries for a query Id. For example, if the granularity specification for invalidation of the client-side query cache contents at the Database Change Notification Module <b>210</b> is at table (object) level, then the below queries may allow for sharing the same query Id: <tables id="TABLE-US-00001" num="1"><table frame="none" colsep="0" rowsep="0"><tgroup align="left" colsep="0" rowsep="0" cols="2"><colspec colname="OFFSET" colwidth="35PT" align="left" /><colspec colname="1" colwidth="182PT" align="left" /><thead><row><entry /><entry /></row><row><entry /><entry namest="OFFSET" nameend="1" align="center" rowsep="1" /></row></thead><tbody valign="top"><row><entry /><entry>Query-1: select column1 from table-A</entry></row><row><entry /><entry>Query-2: select * from table-A</entry></row><row><entry /><entry>Query-3: select column2, column5 from table-A.</entry></row><row><entry /><entry namest="OFFSET" nameend="1" align="center" rowsep="1" /></row></tbody></tgroup></table></tables><br /> The query Id may be used by the Database Change Notification Module <b>210</b> to associate changes in the Database with result sets during query registration.

0051Following registration of the query, <figref idref="DRAWINGS">FIG. 9</figref> is a flowchart for the process of the Database Change Notification Module generation of the Cache Invalidations. The Database Change Notification Module <b>210</b> sets the Out Snapshot to a latest snapshot of the database taken after processing the last client request (<b>900</b>). The Database Change Notification Module <b>210</b> adds Query Ids for the Client <b>200</b> that have changed (if any) between the In Snapshot and the Out Snapshot, representing the latest interaction that the Client <b>200</b> had with the Database, to the set of Invalid Query Ids (<b>902</b>). In one or more embodiments, the invalid query ids are determined by comparing the snapshot of the database at the time the registered query is invalidated to the In Snapshot and Out Snapshot. If the snapshot at invalidation of the query is greater than In Snapshot and less then or equal to Out Snapshot, it is placed in the list of Invalid Query Ids. Registered queries whose query Ids are not in the list of invalid query ids are guaranteed to be consistent as of the Out Snapshot. The set of invalid query ids represent the query ids that have query results that are invalid in accordance with the granularity specified during registration with the Database Change Notification Module <b>210</b>. Next, the Database Change Notification Module <b>210</b> adds dirtied Query Ids for the session (if any) due to the uncommitted changes to the Dirty Query Ids (<b>904</b>). The set of dirty query ids are the query ids for the cached queries affected by the uncommitted changes made by the current session and require the session to rely on query results returned by the server and not use the locally cached results. The Database Change Notification Module <b>210</b> completes the generation of Cache Invalidations by setting the Cache Invalidations to the sets of Dirty Query Ids and Invalid Query Ids (<b>906</b>). Next, the Out Snapshot is returned by the Database Change Notification Module (<b>908</b>). Continuing with <figref idref="DRAWINGS">FIG. 7</figref>, the Database Server <b>206</b> returns the Results, the Cache Invalidations (if any) and the Out Snapshot to the Database API <b>204</b> (<b>710</b>). In one or more embodiments, if the database request was for a query that is cacheworthy, then the query id and the query execution snapshot is returned.

0052In one or more embodiments of a consistent client-side cache, the Client <b>200</b> may receive early invalidations for database query results that have not been received by the Client <b>200</b> from the Database Server <b>206</b>. For example, the Client <b>200</b> may not receive the results from a database request for a long running query or a query with results remaining in the network queue before sending a later request to the Database Server <b>206</b> and receiving a response from the Database Server <b>206</b> that would invalidate the earlier query whose results are not yet received. In such cases, the response to the later request may return from the Database Server <b>206</b> with results that indicate the cached results from the earlier request should be invalidated before the results from the earlier request are cached or even received by the client-side cache, hence an early invalidation occurs. When the results from the earlier request are received by the client-side cache, the client-side cache will cache the results made invalid by the later request without recognizing the invalidation of the results indicated in the response for the later request.

0053One approach for handling early invalidations is to retain the invalidations (e.g. a query id) and a snapshot of the database at a creation time of the invalidation for the query id. With the retention of the invalidations and the corresponding snapshots, the Cache Manger <b>214</b> may not cache the results of the earlier request in the above example because the snapshot associated with the results from the earlier request will be earlier in time than the snapshot associated with the invalid query id. To implement such an approach, an embodiment may return a query id, a result set and a snapshot of the database at execution of the query from the Database Server <b>206</b> to the Client <b>200</b> with the Database API <b>204</b>. If the query involves a cacheworthy query, then the Cache Manager <b>214</b> of the Client <b>200</b> will ensure that the results are not cached if the snapshot for execution of the query associated with a query id is earlier in time than the snapshot for the invalidation of the same query id. In an approach for handling early invalidations, the consistent client-side cache may rely on multiple Database Server <b>206</b> connections such that the early and later query requests and responses in the above example may use more than one database connection. Garbage collection, memory management, may be implemented to discard (e.g. free, delete allocated space, etc.) the invalidation retained for the query id when the query id invalidation snapshot is older than all of the open sessions <b>202</b> in Client <b>200</b> last snapshot communicated, as described in detail in <figref idref="DRAWINGS">FIG. 6</figref>, to the Database Server <b>206</b>.

0054In one or more embodiments, network problems may result in a delayed response from the Database Server <b>206</b> for the Client <b>200</b> that includes an invalidation for a query and cause performance issues with the client-side cache. For example, prior to the arrival of the delayed response to the Client <b>200</b>, the Client <b>200</b> may make an intermediate database request and receive a response from the Database Server <b>206</b> for the intermediate request that includes a duplicate invalidation of the invalidation in the delayed response or a newer result set for a query designated as invalid in the delayed response. Upon receipt of the invalidation in the delayed response, the Cache Manager <b>214</b> either invalidate a newer result set or unnecessarily invalidate a query that has already been invalidated. One approach to avoid later duplicate invalidations is to retain the database invalidation snapshot associated with the invalid query id and compare it to the snapshot associated with the cached result set. With this approach, if the snapshot associated with the invalidation is older than the snapshot associated with the cached results, then the invalidation is discarded.

0055An embodiment will now be described for handling a Client <b>200</b> that has connections to multiple databases. If Client <b>200</b> that is supported by a Client-side Cache <b>212</b> is simultaneously connected to multiple Database Servers <b>206</b>, it may need to differentiate between result sets from the different databases where the SQL text or user identification is similar. In one or more embodiments, a unique database identifier may be used to differentiate between the result sets from different databases. Optionally, a unique database identifier may include any of the following: database name, a database creation timestamp, a database startup time (i.e. every time the database is shutdown and re-started, there is a new timestamp) and/or any globally unique id. In another embodiment, a database may provide a database Id that is unique to that database. In one or more embodiments, a unique database identifier can be sent to the client during the connection (i.e. Session <b>202</b>) establishment. The unique database identifier can be used to differentiate a result set from the result sets from other databases. In one or more embodiment, it may be beneficial to physically separate the Client-side Cache <b>212</b> for each Database Server <b>206</b>. For example, all Sessions <b>202</b> having a common database identifier could use a different cache from the sessions having a different database identifier. Optionally, a Client-side Cache <b>212</b> could be shared and the unique database identifier may be used to identify a cached result set (e.g. part of compile key or runtime key).

0056An embodiment will now be described for multiple Clients <b>200</b> supported by Client-side Caches <b>212</b> that are connected to the same Database Server <b>206</b>. The Database Server <b>206</b> may need to differentiate between these Client-side Caches <b>212</b> in the implementation of the Client-side Caches <b>212</b>. In one or more embodiments, differentiating Client-side Caches <b>212</b> allows the Database Change Notification Module <b>210</b> to generate invalidations for each Client-side Caches <b>212</b>. Embodiments may use the query id to identify both the query and the Client <b>200</b> that requested the query. Some embodiments may use a separate cache Id to distinguish between the Client-side Caches <b>212</b>. In one or more embodiments, a cache Id can be a sequence number that is incremented for every client cache. The most recent cache Id may be stored persistently to ensure that the cache Id will remain after a restart of the database.

0057An embodiment will now be described for ensuring the completeness of cache invalidations in a clustered database approach. Those skilled in the art will appreciate that the approach may be used for a non-clustered database. A clustered configuration may have several nodes or instances which may share the same physical database. There may be a messaging capability between the instances enabling them to communicate information with each other. Within each instance, there may be a multi-process model in which transactions or queries can be executed concurrently in different server processes.

0058Database transactions executed against the database may be assigned a Commit Snapshot upon commit of a transaction. Each transaction may have its own database wide unique transaction id and the Commit Snapshot is typically recorded in persistent journals (e.g. a transaction table) atomically with the commit. It is possible with a transaction id to read the corresponding transaction table and retrieve the transaction Commit Snapshot (i.e. Commit Snapshot). In general, even if the Commit Snapshot cannot be accurately determined, it may be possible to determine an upper bound on the Commit Snapshot. Queries executed against the database may pick up a consistent Snapshot i.e. the query result set may be guaranteed to contain the effects of all transactions that have a Commit Snapshot less than or equal to the Query Snapshot and no others. Database instances may have their own notion of Lamport Snapshot, which may be a higher than or equal to the Commit Snapshots of all transactions known to the database instance. The Lamport Snapshot may reside in the shared memory of the database instance. Transactions upon commit are assigned a Commit Snapshot which is guaranteed to be higher than the Lamport Snapshot prior to commit. Events on the database server which are sequenced in time i.e. one after the other will see an increasing Lamport Snapshot.

0059The change notification infrastructure returns all invalidations generated by transactions with Commit Snapshot higher than the In Snapshot and Commit Snapshot less than or equal to the Out Snapshot. The Out Snapshot is assigned by the change notification module. These invalidations may be referred to as “inband invalidations” since they are returned synchronously upon every round trip from the client cache as opposed to being delivered asynchronously by the change notification module. The Out Snapshot may have the following properties: (1) higher than or equal to the Commit Snapshot of all changes done in the same round trip, (2) at least as high as the Lamport Snapshot on the instance so that in effect subsequent queries against the client cache can see the most current data, and (3) completeness of invalidations. Completeness of invalidations may be the set of invalidations returned from this module within the Snapshot range (i.e. between the In Snapshot and Out Snapshot) that are complete. If a transaction committed and caused a query id to change and the transaction Commit Snapshot was less than the Out Snapshot, then it must be included in the set. If a transaction commits in the future and caused a query result set to change and was not included in the returned set, it must commit at a Snapshot higher than the Out Snapshot. In one or more embodiments, requests for inband invalidations can be submitted in a process which is independent of the process which performed the transaction commit and caused a query result set to change. It may be on an instance which is different from the instance on which the transaction commit was performed.

0060Prior to commit, the Database Change Notification Module can determine a set of query ids that can be invalidated as a result of the changes done within the transaction. These invalidations (e.g. a list of query Ids) are tagged with the transaction id and recorded in the shared memory of the instance that generated them and synchronously broadcast to remote instances using the inter-instance messaging. At the end of the broadcast, each instance of the database has a copy of the (query id-transaction id) tuples. It may be necessary to publish these invalidations prior to commit. When a request for inband notifications arrives on an instance, the Out Snapshot is picked to be the Lamport Snapshot at the end of the call. This may ensure that it sees all changes done within the call and all recent transaction commits known to this instance. The instance on which the request is generated already knows the candidate set of invalidations (query Id, Transaction Id) tuples which is in its own shared memory. However, some of these may have Commit Snapshots higher than the Out Snapshot so it may not be included in the answer returned to the client-side cache. In order to determine which invalidations to include, the transaction tables may be consulted to obtain the Commit Snapshot of the transactions. If the Commit Snapshot is lower than or equal to the Out Snapshot, the corresponding query id is included or it is saved for later. Consulting the transaction table involves acquiring a short duration read lock (i.e. referred to as a pin) on the transaction table. A transaction which was active (perhaps on a different instance) at the time the request was being processed is guaranteed to commit with a Commit Snapshot higher than the Out Snapshot thereby guaranteeing the completeness of the invalidations because upon commit this transaction will acquire a short duration write lock to record its own Commit Snapshot. Because of the Lamport sequencing property described above, the Commit Snapshot of this transaction will be higher than the Out Snapshot thereby ensuring the completeness of invalidations returned in the previous result. The server may cleanup invalidations from its shared memory when the In Snapshot of the client cache has advanced beyond the Commit Snapshot of the transaction that generated a given invalidation. In this case, it is known that the client cache already has knowledge of this invalidation and hence it can be safely reclaimed.

0061An embodiment will now be described for handling normal client termination. When a Client-side Cache <b>212</b> process exits, the Database API <b>204</b> may be used to disconnect or terminate a Session <b>202</b> and halt the Database Change Notification Module <b>210</b> from generating a set of cache invalidations for registered queries that are pertinent to the Client <b>200</b> and/or the Session <b>202</b>. In one or more embodiments, if multiple sessions are sharing the Client-side Cache <b>212</b>, a reference count of the number of Sessions <b>202</b> relying on the Client-side Cache <b>212</b> can be maintained. If no sessions remain according to the reference count, then the Database Change Notification Module <b>210</b> may be prevented from generating a set of cache invalidations for registered queries that are pertinent to the Session <b>202</b> and garbage collection/cleanup may be performed. In one or more embodiments, Cache Ids can be part of a session state on the Database Server <b>206</b>, and pertinent to the Client-side Cache <b>212</b> related persistent database tables. If the session state goes away as a result of client normal termination, a periodic database cleanup action may be used to find the cache Ids in the persistent tables. A cleanup/garbage collection action may be performed to free up the cache state including resources held in the Database Change Notification Module <b>210</b> used for generating a set of cache invalidations for registered queries that are pertinent to the Client <b>200</b> and/or the Session <b>202</b>. In one or more embodiments, the Client-side Cache <b>212</b> may be a daemon process on the same or different machine, and this cache daemon may continue to be alive until the database is up, or periodically refresh the cached result sets or periodically re-start itself.

0062An embodiment will now be described for handling abnormal client termination. If a Client-side Cache <b>212</b> process dies before calling a cleanup/garbage collection action to free up resources including resources held in the Database Change Notification Module <b>210</b>, it may impact the Database Server <b>206</b>. In one or more embodiments, the Database Server <b>206</b> may perform periodic actions to check for the need to perform such cleanup action. The cleanup action may include freeing in-memory resources and persistent resources, for the abnormally terminated Client-side Cache <b>212</b>.

0063In one embodiment, the cache Ids can be part of session state on the Database Server <b>206</b>, and also in the client cache related persistent database tables. If the session state goes away as a result of client abnormal termination, the periodic database cleanup action may find the cache Ids in the persistent tables but not in the server session state. A cleanup/garbage collection action may be performed to free up the cache state at the server including resources held in the Database Change Notification Module <b>210</b>.

0064An embodiment will now be described for handling duplicate result sets. During query execution for the Client <b>200</b>, a duplicate result set may exist in the Client-side Cache <b>212</b> that is not invalidated. The caching for this result may be complete or incomplete. If caching is incomplete, the Cache Manager <b>214</b> in on or more embodiments may force this query execution to not create another cached result set on Client <b>200</b>, while getting result set from Database Server <b>206</b>. If caching is complete, it is not necessary for Cache Manager <b>214</b> to create another result set in the Client-side Cache <b>212</b>. In one or more embodiments, an algorithm for retrieval of query execution results does not include invalid result sets in client cache and may ensure subsequent query executions will get result set from Database Server <b>206</b> to create a new result set at the Client <b>200</b>.

0065An embodiment is now described for handling changing environment settings At any point, the Client <b>200</b> may change environment or session settings that may affect the result sets cached on Client <b>200</b>. Database Server <b>206</b> calls made by same or different Clients <b>200</b> may also change environment settings that may affect result sets cached on various Client-side Cache <b>212</b>. In one or more embodiments, the Client-side Cache <b>212</b> detects such changes in environment settings on its next Database Request (<b>204</b>) to the Database Server <b>206</b>. The Database Server <b>206</b> request in one or more embodiments may return a new environment state, as piggyback. By always including the environment settings as part of runtime key computation, the Client-side Cache <b>212</b> may ensure that the query executions with different environment or session settings do not share result sets. In one or more embodiments, alteration of any environment or session state may affect query result sets that cant invalidate all result sets in the Client-side Cache <b>212</b>. Examples of environment or session setting include language setting, time zone settings, and result-set output format.

System Architecture

0066The execution of the sequences of instructions required to practice the embodiments may be performed by a computer system <b>1400</b> as shown in <figref idref="DRAWINGS">FIG. 10</figref>. In an embodiment, execution of the sequences of instructions is performed by a single computer system <b>1400</b>. According to other embodiments, two or more computer systems <b>1400</b> coupled by a communication link <b>1415</b> may perform the sequence of instructions in coordination with one another. Although a description of only one computer system <b>1400</b> will be presented below, however, it should be understood that any number of computer systems <b>1400</b> may be employed to practice the embodiments.

0067A computer system <b>1400</b> according to an embodiment will now be described with reference to <figref idref="DRAWINGS">FIG. 10</figref>, which is a block diagram of the functional components of a computer system <b>1400</b>. As used herein, the term computer system <b>1400</b> is broadly used to describe any computing device that can store and independently run one or more programs.

0068Each computer system <b>1400</b> may include a communication interface <b>1414</b> coupled to the bus <b>1406</b>. The communication interface <b>1414</b> provides two-way communication between computer systems <b>1400</b>. The communication interface <b>1414</b> of a respective computer system <b>1400</b> transmits and receives electrical, electromagnetic or optical signals, that include data streams representing various types of signal information, e.g., instructions, messages and data. A communication link <b>1415</b> links one computer system <b>1400</b> with another computer system <b>1400</b>. For example, the communication link <b>1415</b> may be a LAN, in which case the communication interface <b>1414</b> may be a LAN card, or the communication link <b>1415</b> may be a PSTN, in which case the communication interface <b>1414</b> may be an integrated services digital network (ISDN) card or a modem, or the communication link <b>1415</b> may be the Internet, in which case the communication interface <b>1414</b> may be a dial-up, cable or wireless modem.

0069A computer system <b>1400</b> may transmit and receive messages, data, and instructions, including program, i.e., application, code, through its respective communication link <b>1415</b> and communication interface <b>1414</b>. Received program code may be executed by the respective processor(s) <b>1407</b> as it is received, and/or stored in the storage device <b>1410</b>, or other associated non-volatile media, for later execution.

0070In an embodiment, the computer system <b>1400</b> operates in conjunction with a data storage system <b>1431</b>, e.g., a data storage system <b>1431</b> that contains a database <b>1432</b> that is readily accessible by the computer system <b>1400</b>. The computer system <b>1400</b> communicates with the data storage system <b>1431</b> through a data interface <b>1433</b>. A data interface <b>1433</b>, which is coupled to the bus <b>1406</b>, transmits and receives electrical, electromagnetic or optical signals, that include data streams representing various types of signal information, e.g., instructions, messages and data. In embodiments, the functions of the data interface <b>1433</b> may be performed by the communication interface <b>1414</b>.

0071Computer system <b>1400</b> includes a bus <b>1406</b> or other communication mechanism for communicating instructions, messages and data, collectively, information, and one or more processors <b>1407</b> coupled with the bus <b>1406</b> for processing information. Computer system <b>1400</b> also includes a main memory <b>1408</b>, such as a random access memory (RAM) or other dynamic storage device, coupled to the bus <b>1406</b> for storing dynamic data and instructions to be executed by the processor(s) <b>1407</b>. The main memory <b>1408</b> also may be used for storing temporary data, i.e., variables, or other intermediate information during execution of instructions by the processor(s) <b>1407</b>.

0072The computer system <b>1400</b> may further include a read only memory (ROM) <b>1409</b> or other static storage device coupled to the bus <b>1406</b> for storing static data and instructions for the processor(s) <b>1407</b>. A storage device <b>1410</b>, such as a magnetic disk or optical disk, may also be provided and coupled to the bus <b>1406</b> for storing data and instructions for the processor(s) <b>1407</b>.

0073A computer system <b>1400</b> may be coupled via the bus <b>1406</b> to a display device <b>1411</b>, such as, but not limited to, a cathode ray tube (CRT), for displaying information to a user. An input device <b>1412</b>, e.g., alphanumeric and other keys, is coupled to the bus <b>1406</b> for communicating information and command selections to the processor(s) <b>1407</b>.

0074According to one embodiment, an individual computer system <b>1400</b> performs specific operations by their respective processor(s) <b>1407</b> executing one or more sequences of one or more instructions contained in the main memory <b>1408</b>. Such instructions may be read into the main memory <b>1408</b> from another computer-usable medium, such as the ROM <b>1409</b> or the storage device <b>1410</b>. Execution of the sequences of instructions contained in the main memory <b>1408</b> causes the processor(s) <b>1407</b> to perform the processes described herein. In alternative embodiments, hard-wired circuitry may be used in place of or in combination with software instructions. Thus, embodiments are not limited to any specific combination of hardware circuitry and/or software.

0075The term “computer-usable medium,” as used herein, refers to any medium that provides information or is usable by the processor(s) <b>1407</b>. Such a medium may take many forms, including, but not limited to, non-volatile, volatile and transmission media. Non-volatile media, i.e., media that can retain information in the absence of power, includes the ROM <b>1409</b>, CD ROM, magnetic tape, and magnetic discs. Volatile media, i.e., media that can not retain information in the absence of power, includes the main memory <b>1408</b>. Transmission media includes coaxial cables, copper wire and fiber optics, including the wires that comprise the bus <b>1406</b>. Logic refers to software, hardware or any combination of software and hardware.

0076In the foregoing specification, the embodiments have been described with reference to specific elements thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the embodiments. For example, the reader is to understand that the specific ordering and combination of process actions shown in the process flow diagrams described herein is merely illustrative, and that using different or additional process actions, or a different combination or ordering of process actions can be used to enact the embodiments. The specification and drawings are, accordingly, to be regarded in an illustrative rather than restrictive sense.