Analytics – The key priority for CIO spending
Fri 13 May 2016
When Gartner surveyed 3,000 CIOs, their top spending priority wasn’t cloud, it wasn’t security, it wasn’t even data protection. No, it was Analytics. No longer is just reporting sufficient, CIOs want – and need – insight that will help drive business advantage. From providing better business intelligence to enhanced storage performance and cost, no wonder analytics are top of the spending priority list. This article looks at an area of IT infrastructure – the virtualised data centre – that it’s not enough to just see what is happening, CIOs want to know why things are happening, the larger impact and what remedial action needs to take place.
Drilling down into a virtualised data centre shouldn’t be a mystery. But without the right management platform it can be hard, nigh on impossible at times, to know what is going on in your virtualised data centre.
Here are 6 things you probably don’t know about the VMs in your data centre but can cause a domino effect on other parts of the infrastructure if you don’t have the right level of visibility.
Workloads are constantly shifting
Workloads constantly change based on VM, application and infrastructure requirements. Tools that sample data every 5 or 10 minutes are less accurate than those that continuously sample data. This is critical for correctly sizing persistent storage, caching tiers, data replication and more.
Reads and writes behave differently
All storage platforms handle reads differently from writes. Understanding your application read/write mix is therefore critical for managing storage performance and capacity. In addition, some storage platforms are better at reads and/or writes than others. Therefore, understanding this mix is also critical when making storage purchasing decisions.
Block size matters
Often considered the most hidden and misunderstood performance metrics in a data centre, block sizes have a profound impact on the performance of storage infrastructure. VMs, and the applications that run in them, will often suffer when they use block sizes that the storage system cannot adequately handle.
The big picture
It is not enough to look at individual metrics on their own. One must be able to correlate these measurements across data centers for holistic storage management. For example, when managing storage you need visibility into how IOPs, latency, and throughput all collectively impace the performance of every VM.
Location is everything
Management solutions that depend on VMware vCenter rarely correlate VM data with storage data, limiting the ability to address storage challenges that impact VM performance. Storage array vendors’ management products are not always accurate when assessing VM performance because so many things outside of the array can impact VM latency. To accurately collect and correlate VM and storage intelligence, a management tool is best placed inside the hypervisor kernel.
Access patterns affect performance
All data centres have a mix of sequential and random workloads. Performance of the latter can suffer when sequential workloads increase congestion, fill up caches and user large block sizes. To properly design and manage storage, you need to understand your data mix and plan for different access patterns.
So, ensure you can provide your CIO the right information to enable correct decisions to be made.
In a virtualised data centre you want to be able to have faster VM troubleshooting via real-time analytics, lower IT costs through better storage design and more efficient hardware procurement and, finally, optimised application performance with better infrastructure planning
With the right infrastructure analytics you can turn visibility into knowledge, lower IT costs and improve business operations in the process.