SSDs and Application Performance: What’s the connect?
Before I even start getting into
the details of SSD and application performance, a foreword about what is the intent
behind writing this post. This post will mostly run you through understanding
of performance problems in data center that use magnetic disks, will touch base
on major factors that govern application performance.
Is network latency the culprit?
Applications are slowed by
network latencies: is now quite a known fact. Apps that run over a network are
hugely impacted by network and storage levels too. This typically is followed
by I/O bottlenecks popping up as more and more users login to make the
application work even harder.
Traditional hard disk drive‐based
systems incur performance penalty because of the high access times of hard disk
drives. However, Solid state disks resolve this issue primarily with low access
times and high bandwidth.
Access Time = Command Overhead Time + Seek Time
+ Settle Time + Latency
Direct impact on Application performance
Greatest of all implications posed
are random workloads, sequential workload, and the ones that are read and write
intensive.
-
Low IOPS (I/O per second) and high I/O wait time based on high access times for HDD based storage access
- Wasted server infrastructure caused by businesses responding to performance problems by buying more processors and more servers
- Long application response times as wait time build up
- Decreased user productivity and decreased customer satisfaction caused by application response time delays
- Long running batch jobs caused by increasing I/O wait time
- Shorter maintenance windows and failure to meet service level agreements due to long running batch jobs
What
role has SSD to play?
Solid state disks are
non‐volatile storage devices that use flash as the primary storage media. In
the case of Flash SSD, non‐volatility is achieved due to the inherent non-volatility
of Flash memory. Factors that attribute to enhanced application performance in
light of SSDs are:
High I/Os per second (IOPS): Most SSDs have access times below 15
microseconds, the best access times of any storage platform. This enhances
application performance in terms of reduced I/O bottlenecks and increased user
concurrency.
High bandwidth: Enterprise SSDs can support as much as 4.5 GB per second
of random data throughput. They are committed to delivering benefits such as
lower power consumption, faster response times, increased customer satisfaction
and higher profits.
To know more email: blog@calsoftinc.com
