In midsize or larger IT infrastructures you will typically find that there are several storage systems for dedicated use cases. That is why storage professionals are continuously involved in consolidation projects. Here are five recommendations that may help you when you start your next storage consolidation project.
1. Too big is too expensive!
A typical phenomenon in storage consolidation projects is that the systems purchased are bigger than necessary. Larger storage systems tend to have higher prices per capacity than smaller ones. You will end up wasting money if you plan for too much headroom. A better tactic to deal with such a scenario will be to consider storage system families, allowing you to upgrade smaller models to bigger ones.
2. Performance is key – but beware of performance marketing!
The more data you have, the more performance you will need to be able to read and write data simultaneously. However, there is a lot of performance marketing material out there, using unrealistic values. You should therefore use official benchmarks published by the Storage Performance Council (SPC) in order to select the right product. Do not be guided solely by IOPS figures, but also by how much data the system can handle per second (bandwidth). Thereafter, the value that really counts is the response time under full workload. You will find this in the small print in-official SPC benchmarks. If a vendor does not publish official benchmarks, then better ask for real performance tests before you buy.
3. All-flash is all-right now
Until now, even an already consolidated storage system may require ongoing tuning efforts when a large number of applications concurrently access data with diverse access patterns. The low latency of all-flash storage no longer requires so much optimization. At the same time, maintenance work will decline as, unlike hard disks, SSD’s need to be replaced very rarely. Being a “disk jockey” is no longer part of the administrator’s job description. SSDs also save a lot of power and space. Imagine that you can store more than 6 Petabytes in a single rack. It is true that the price per capacity for SSD’s is still higher than for fast SAS disks, but you will save a lot after purchase. So, take a look at the total costs. You can also expect that this price gap will melt away within the lifetime of your next storage system. So, consider all-flash now!
4. Diversity likes disks!
A hybrid storage system with SSD’s and slow spinning large capacity disks is ideal, when you plan to host data for performance-hungry applications like virtualization, databases or OLTP alongside unstructured data, online archives and other “luke-warm” data in a single system. This allows you to balance speed, capacity and costs in one system. However, it also means a return to performance tuning efforts. The remedy is to be found in quality of service management features, allowing you to prioritize performance allocation per data volume in an automated way. This becomes even more effective if you combine it with automatic storage tiering.
5. Beware of the dark side of consolidation
If you put more eggs in one basket, the negative impact in the event of a disaster is higher. Hence, every storage consolidation project should include disaster recovery concepts. Your storage system should at least mirror or replicate data on a second system. It is even better if a transparent failover can automatically be triggered once a storage system completely fails. Ideally your storage vendor offers capabilities to mix and merge system sizes and hybrid storage with all-flash storage in DR configurations. So, if not all of your data is mission-critical you can use a smaller storage system as stand-in system, for example. Or you can use a less costly hybrid storage system as a failover system for your all-flash array. Both can be a great cost savers for fault tolerance.