Current Filter: Storage>>>>>Technology Focus>

   

These shifts in expectations mean customer-facing or key operational platforms must provide a consistent user experience at all times. This is an absolute necessity and can provide an organisation with a measurable competitive advantage.

CONSISTENCY IS KING
The reason I use the word 'consistency' rather than speed is because the highest running speed of a database environment is an unhelpful performance measure. Databases underpinned by traditional Hard Disk Drive (HDD) infrastructures are perfectly capable of providing high speeds at times when traffic levels are average or below. What matters most is how a platform performs at times of peak traffic. From a commercial standpoint, this is when speed matters most, as your organisation is potentially challenging customer loyalty and missing out on more custom.

For example, a ticketing website will experience extreme highs and lows in terms of website traffic. However, given the changing consumer expectations of online transactions, loading times need to be as fast on the day 'One Direction' tickets go on sale as they are on a relatively quiet day. A corporate example is the required operational speed of a trading platform the day a fast-growing technology company floats on the stock market. High speed decisions need to be made by investors and traders based on real-time information.

Any delays in the transmission of this information can result in significant losses for organisations and individuals, as well as - in extreme cases - in breaches of regulatory guidelines.

These types of applications requiring consistent performance across the board need to be based on an infrastructure that provides extremely low latency. Traditional HDD-based infrastructures can only handle a finite amount of I/O processing before they become saturated. Organisations have dealt with this issue of peak traffic and data by over provisioning HDD spindles and, therefore, capacity. This process is known as 'short-stroking'.

This balancing act between I/O performance and capacity has, for many, resulted in inefficient storage capacity provisioning in order to meet peak demand. The intelligent deployment of Flash technologies has dramatically changed this dynamic. SSD technology has orders of magnitude higher read and write I/O capability than HDD.

LESS PATIENCE, MORE FLASH
For the reasons cited above, today, Flash is predominantly reserved for performance and latency sensitive business applications; those applications that cannot afford any inconsistency in terms of speed or performance at any time, including peak periods of usage. However, there are certain applications for which flash technology is complete overkill.

Not all workloads require extreme performance all of the time. Put simply, workloads can be categorised as requiring Extreme Performance, Variable Performance ('Performance when I need it, not when I don't') and Capacity Performance. The skill is in mapping business value needs to the required performance and capacity profile.

Increasing performance demands as a clear business differentiator are becoming more influential in developing the business case for Flash. One clear application of Flash technology is for acceleration of OLTP Database workloads. In these scenarios, it is often found that traditional HDD based architectures present an I/O bottleneck. As I said previously, this is why short-stroking HDDs became popular.

Page   1  2