Intel® Cache Acceleration Software: A Primer

By copying frequently accessed enterprise data onto fast Intel® Solid State Drives, Intel® Cache Acceleration Software (Intel® CAS) delivers striking increases in workload I/O performance: for example, it delivers up to three times more performance on transactional database processing and up to 20 times faster processing of read-intensive business analytics.1

This article helps both business-oriented and technology-focused decision makers streamline the process of assessing the technology.

Enterprise data stores continue to grow dramatically, as does the performance of processors built to create value and intelligence from that data. The Intel® Xeon® processor E7 v2 family, for example, delivers up to double the average performance of previous platforms, and it has three times the memory capacity.2 Data center applications that take advantage of that performance commonly dispatch workloads so quickly that they can become starved of data, constrained by the limitations of conventional hard disk drives (HDDs) based on spinning magnetic disks.

Solid state drives (SDDs) have offered a ready alternative, with superior performance compared to HDDs for the most demanding data center workloads. However, cost considerations have led many budget-conscious enterprises to limit their use of SSDs. Intel CAS can easily deliver enormous performance gains with a relatively minor addition of SSDs, while retaining existing investments in conventional HDDs. At a high level, the solution uses the following mechanism:

  • Identify which data to cache. Intel CAS applies algorithms to assess which data would deliver the most value by being cached; IT policy can also lock specific data into the cache, such as to meet a particular SLA.
  • Copy that data to the SSD. The data is available to the processor far more rapidly from the SSD than it would be from its primary location on the HDD-based main data store.
  • Continually reassess optimal cache usage. Intel CAS dynamically identifies which data will deliver the most benefit by being cached, based on changing workloads.

This article provides an overview of Intel CAS, its implementation, and its benefits to particular environments. In addition to acting as a general introduction to both business and technical considerations, this primer also provides links to additional resources, to help support in-depth assessment of the solution.

Intel CAS at Work: A Closer Look

When Intel CAS is enabled on a server, data is written to the Intel CAS cache (SSD) when it is first read from main storage. The second time the data is read, it is copied to system memory (RAM), and future reads are returned at the high speed of either RAM or the SSD. All changes to that data are written simultaneously to the version in cache and the one in main storage, keeping the two in sync.

To optimize the utilization of the SSD cache, only the active parts of applications are cached. When the cache is full, newly identified “hot” data evicts and replaces older, stale data on an ongoing basis. The following resources provide additional detail about the operation of Intel CAS:

How Intel CAS Fits into Your Environment

No special integration effort is needed to bring Intel CAS into an existing data center environment. It is transparent to users, applications, and storage systems, and no changes are needed to applications or the network environment.

Intel CAS installs directly into the Microsoft Windows Server* or Linux* OS, whether running directly on the server hardware or virtually under a VMware, KVM, Citrix Xen Server*, or Windows Hyper-V environment (including live migration while keeping the cache intact). The following resources help explain how Intel CAS can benefit your data center:

Workloads that Will Benefit Most

Data center operators struggling to manage large amounts of transactional, customer, regulatory, or other data often run up against the limitations of HDD-based storage. Intel CAS and Intel SSDs can work together to help alleviate those limitations specifically in cases where application performance is limited because of slow data access from conventional HDDs, by caching “hot” data that is frequently needed by the processor.

Note that such “I/O-constrained” workloads are distinct from “processor-constrained” ones where applications are waiting for the processor to finish executing instructions. Therefore, it is important to consider whether a given workload is constrained by access to data, the speed with which the processor can work on that data, or something else (such as the availability of sufficient memory). The following resources help end customers identify the workloads that will benefit most from Intel CAS:

Examples of Real-World Results

Enterprise-scale end customers across industries have realized extraordinary results from their implementations of Intel CAS, on workloads ranging from business intelligence to high-end graphics. The following series of case studies captures performance results that were obtained using Intel CAS in conjunction with Intel SSDs on common enterprise workloads:

Those results are also borne out in the lab environment. The following resources report the results of lab testing that is designed to quantify the benefits available from implementing Intel CAS:

Conclusion and Next Steps

Intel CAS provides a simple, cost-effective way of using a modest investment in Intel SSDs to significantly increase the performance of applications that are constrained by slow access to data on conventional HDDs. Because Intel CAS was built specifically to take advantage of the capabilities of Intel SSDs, the combination of Intel CAS and Intel SSDs is recommended. The following resources enable potential end customers to advance their assessment of the technology toward a proof of concept or full implementation:

1    Software and workloads used in performance tests may have been optimized for performance only on Intel® microprocessors. Performance tests, such as SYSmark* and MobileMark*, are measured using specific computer systems, components, software, operations, and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. Based on the following configuration: Intel® Server Board 2600CO (Copper Pass); Intel® Xeon® processor E5-2680 (2.7GHz), 32GB DDR2/1333 memory; Microsoft Windows* 2008R2 SP1, Intel® CAS 2.0 release candidate 1; I/O meter 10.22.2009 ; 4K random read test; 32-queue depth; 800GB Intel® SSD 910 series, Intel® RAID RS25AB080 with MR54p1 firmware; 8 x 10K SAS HDD in a RAID0 array with MR54p1 firmware; and 8 x 10K SAS HDD in a RAID0 array. For more information go to http://www.intel.com/performance.

2   See the Intel Xeon processor E7-8800/4800/2800 v2 product brief, “The Foundation for Better Business Intelligence,” at http://www.intel.com/content/www/us/en/processors/xeon/xeon-e7-v2-family-brief.html

Para obter informações mais completas sobre otimizações do compilador, consulte nosso aviso de otimização.