Proposing EMC FAST Cache and FAST VP – Are you Sure ?

Folks, I see a lot of time people just propose FAST Cache and VP without actually doing a deep dive into it. I mean if you don’t know whether your customer actually requires it, it would not be a wise idea to propose these solutions.

So let me tell you the situation when you should not propose these solutions.

  • If your customer data has no skew, that means that I/Os to a LUN are not evenly distributed over the entire data area. Some areas are very busy, while others may be accessed very infrequently. An example of skew is when 5% of the customer data generates 95% of all I/Os.
  • When your customer cannot tolerate a false positive. A false positive means that critical data was placed on a slower tier when it was needed on the fastest tier.
  • When your customer has unrealistic expectations. For example, customer data has traditionally been spread across 100s of 15K drives, and the goal is to replace this with 2 Flash drives in FAST Cache.
  • Finally, in cases where there is skew, do NOT under size the FLASH tier. If 5% of all data is responsible for 95% of all IOs, have 1% at least in FAST Cache and then 5‐7% as a Flash tier.
  • When your customer is running sequencial workloads on LUNs, especially where I/O sizes are small. In this case an example of Backup workload will impact their storage performance significantly.

Now what you should remember and facts you should take into considerations are:

FAST VP should be given time to learn the environment. This is an important expectation to set with customers, especially important when data is migrated from a high disk count
environment. There is a huge initial difference between a 100% 15K system and a 5%‐20%‐75% VNX with FAST VP. It can take FAST VP several days to learn which data is hot and get these slices moved to the higher tiers.

This warm up phase for FAST Cache can take from several minutes to several hours, and this should be taken into account.

For improved data availability, allocate drives that make up a FAST Cache RAID 1 pair to “different” backend buses. This may also have a positive effect on FAST Cache performance. Note that, as a general rule, Flash drives should be spread across multiple buses when the number of Flash drives in use demands it. More than 4 Flash drives can saturate a backend bus, so consider drive placement carefully when using FAST Cache, FAST VP or even Flash only storage pools.

About Prasenjit Sarkar

Prasenjit Sarkar is a Product Manager at Oracle for their Public Cloud with primary focus on Cloud Strategy, Oracle Openstack, PaaS, Cloud Native Applications and API Platform. His primary focus is driving Oracle’s Cloud Computing business with commercial and public sector customers; helping to shape and deliver on a strategy to build broad use of Oracle’s Infrastructure as a Service (IaaS) and Platform as a Service (PaaS) offerings such as Compute, Storage, Java as a Service, and Database as a Service. He is also responsible for developing public/private cloud integration strategies, customer’s Cloud Computing architecture vision, future state architectures, and implementable architecture roadmaps in the context of the public, private, and hybrid cloud computing solutions Oracle can offer.

One thought on “Proposing EMC FAST Cache and FAST VP – Are you Sure ?

  1. I just received a new VNX 5500 from EMC with 17 100GB Flash Drives. Guess how they were laid out? 7 on the vault DAE, 9 in another DAE on the same bus and 1 in another DAE.

    Luckily I’m aware of the bus saturation issue, but I surprised it was shipped in such a non optimal configuration.