Intel® solid state drives (SSDs) are increasingly used in the enterprise and high-performance computing environments to accelerate applications for improved performance. Typical applications that are best suited for SSD acceleration include virtualization, database, and big data analytics.
A great cost-efficient model for memory-constrained servers is the ability to leverage SSDs for caching. In private cloud hosts that utilize memory over-subscription, the hypervisor reclaims memory by swapping memory pages to the host cache. It is important to note this is a memory page swap improvement feature, not a storage I/O feature.
Data stores created on SSDs can be used to allocate space for host cache. The host reserves a certain amount of space for swapping to host cache. The host cache is made of files on an SSD that the hypervisor uses as a write-back cache for virtual machine swap files. The cache is shared by all virtual machines running on the host.
To use local SSDs for host caching with over-committing the host server memory, it is important to calculate the amount of cache needed to keep these pages in the cache. Based on our experience, we choose an SSD equal to the size of the host’s memory. For example, in a host with 256 GB of RAM we would use a 256-GB SSD.
Note that each host will need to have its own host cache assigned; therefore, it makes sense to use a local SSD instead of a network storage volume. Generally speaking, network storage volumes are much more expensive than local volumes and using local volumes removes the latency caused by the storage network. Compared to the latency of an Intel SSD (less than 100 microseconds), network latency can be significant.
Configuring the hypervisor to “swap to host cache” with Intel SSDs can control the application performance degradation that is caused by swapping—often even under significant memory pressure. This helps increase virtual machine density, maintain performance at a given service level agreement, and reduce cost per virtual machine.
From a business perspective, the cost per virtual machine can be reduced by configuring host base caching, resulting in both CapEx and OpEx savings. Dense private clouds have a more efficient footprint, which intuitively should lead to a lower TCO. This smaller data center footprint can reduce TCO in areas of data center power, cooling, storage, and network connectivity costs. Host base caching reduces traffic on storage networks and storage I/O contention, which can delay the need for network upgrades.
Additionally, increased virtual machine density per cloud host can reduce costs related to the license and maintenance costs of software such as OS, applications, middleware, security products, backup and restore, and manageability.
To recap: advantages of Intel SSDs include reduced power usage, faster data access, and higher reliability. Intel SSDs used for local host swap cache can provide a significant increase in performance when contrasted to relying on network-based storage.
To get more details on Intel IT and our use of SSDs see our white paper, “Increasing VM Densities and Performance in Private Cloud Hosts with an SSD” or visit intel.com/IT.