SuperCache 2000 How-to

Sybase Server

The problem: Sybase performance, like that of similar services, can quickly degrade under heavier loads. During times of peak demand, application processes that lock tables may block other processes for unacceptably long periods. In such situations, the offending process may have to be terminated. Worse still, under very heavy loads a point of deadlock may be reached requiring that the database engine be stopped and restarted.

Even when the server is under moderate loads, data- and query-intensive operations such as reporting may suffer extended latencies, resulting in user waits of many minutes instead of seconds for an operation to complete. Code optimizations improve performance to a degree, but frequently the single most important factor is the storage subsystem.

The solution: SuperCache can entirely eliminate, or at least greatly mitigate, these issues. It inserts itself into the disk I/O path at the block level, and caches the most active blocks of data on the logical disks selected. When deferred-writing is enabled, SuperCache dramatically improves server performance, often reducing the duration of application operations by an order of magnitude or more. Extended periods of process blocking and deadlocks in the Sybase engine disappear.

How it's implemented: For maximum benefit, ÍhotÍ data as well as the page file should be located on a separate logical disk (disk partition or disk volume), which in turn should be cached by SuperCache. SuperCache trades CPU utilization and physical memory for disk I/O bandwidth. The larger the cache, the better the performance.

Messaging and collaboration servers, such as IBM's Lotus Domino server and Microsoft's Exchange Server, are responsible for managing the flow of various forms of interactive communications among users, greatly enhancing the dynamics of those interactions. However, as the number of users and communications increase, the ability of the server to nimbly respond to user requests diminishes.

The problem: When higher numbers of user interactions occur, the time required for a user to view his or her schedule, contacts, and tasks, or to send or received e-mails, etc., may increase from a fraction of a second to many seconds. Web page renderings may take ten times as long - or even longer - to complete. These latencies begin to erode a user's sense of productivity, detracting from the objective of efficient collaboration.

The cause of these conditions is simple: as the volume of data being retrieved and stored by the server grows, the server's storage subsystem constrains the system's overall performance. Although Lotus Domino and Exchange Server feature internal caches, their effectiveness is reduced as the number of users and I/O transactions increase.

The solution: SuperCache alleviates latency problems by removing the performance bottleneck. It inserts itself into the disk I/O path at the block level, and caches the most active blocks of data on the logical disks selected. When deferred-writing is enabled, SuperCache significantly improves server performance, and dramatically reduces user request latencies, resulting in a much more satisfying and productive user experience.

How it's implemented: For maximum benefit, ÍhotÍ data (such as the stores), and the page file should be located on a separate logical disk (disk partition or disk volume), which in turn should be cached by SuperCache. SuperCache trades CPU utilization and physical memory for disk I/O bandwidth. Generally, the larger the cache, the better the performance. Domino and Exchange users on heavily loaded servers commonly experience latency reductions of ten times or more when SuperCache is deployed.

Whether you run Navision on SQL or Native Navision you can enhance speed and performance dramatically with SuperCacheII.

SuperCache II use patented block-level caching technology to raise system performance far beyond that of the NT file-level cache alone. It breaks the disk I/O bottleneck by using the server's system memory to store the most frequently used data, executing read and write requests in memory rather than on disk. SuperCache is implemented as a storage volume filter driver, residing in the storage stack just above the storage volume class drivers.

How it works: SuperCache intelligently stores frequently used disk data in system memory. It stores the data in an area called its cache. It intercepts requests for disk data and, when possible, fulfills the request using the cache instead of requesting data from the disk. When data comes from or goes to the cache, the performance is similar to a RAM disk - extremely fast.

Why another cache? Doesn't Windows already have a cache? The Windows operating system supports a file-system level cache, the NT file-system cache. However, the flow of data at the file-system level and the flow of data at the block (or storage volume) level are not the same. Each level - the file-system level and the block level - represents a different data flow paradigm, or model. Optimization of flows at the file system level does not optimize flows at the block level. SuperCache enhances disk I/O throughput by optimizing flows at the block-level.

The problem: Lots of files are accessed by tens or hundreds of users, often the same data is accessed, especially when producing large reports.

The solution: SuperCache alleviates latency problems by removing the performance bottleneck. It inserts itself into the disk I/O path at the block level, and caches the most active blocks of data on the logical disks selected. When deferred-writing is enabled, SuperCache significantly improves server performance, and dramatically reduces user request latencies, resulting in a much more satisfying and productive user experience.

How it's implemented: For maximum benefit, ÍhotÍ data (such as the stores), and the page file should be located on a separate logical disk (disk partition or disk volume), which in turn should be cached by SuperCache. SuperCache trades CPU utilization and physical memory for disk I/O bandwidth. Generally, the larger the cache, the better the performance. Microsoft Navision and Axapta servers are heavely loaded and commonly experience latency reductions of ten times or more when SuperCache is deployed. By not using the lazy-write you are ensured maximum performance with 100% data security.

SuperCache ensures:

• Data written to disk immediatly for 100% fail Safe Ramdisk


• Reads are performed through the SuperCache at extreme speed