In this busy world of HyperConverged and Converged Infrastructure offerings. Last week NetApp released a new beast. The FlexPod SF, and it’s exactly that. A converged offering from Cisco and NetApp with SolidFire storage capability running alongside Cisco networking and compute in a validated and reference architecture.
A cool marketing video if anything –
The FlexPod Portfolio
As I said FlexPod isn’t new from both Cisco & NetApp they have been doing it for years. Before I joined Veeam this was my bread and butter from a design and implementation point of view.
Introducing FlexPod SF
It is suggested that the FlexPod SF use case is going to be for the following 3 areas and I will share my opinion on why this will most likely work:
Core Data Centre – Although NetApp are and have branded SolidFire the Next-Generation Data Center answer to storage, there is for a while yet going to be the “traditional” data centre requirements that we love or hate today, granted todays VMware might be tomorrows OpenStack and this is geared up for both to quote some marchtecture from NetApp “Linear Compute and Storage scaling without interruption”
Service Providers – This was the first place I saw SolidFire a few years back in a Data Centre where I was implementing a FlexPod environment, I did some reading up after that and published a 101 on my blog site and I mention there about how QoS engine within the Element OS is great for offering that performance to multi-tenant environments, also now being able to put a policy and automation against the compute and storage just adds more control in those environments.
DevOps – Another area that seems to have come with the SolidFire badge is the “DevOps” buzz, full APIs available and built with APIs first is the new normal but this alone really allows for that true DevOps environment to achieve easy automated deployment and allows for much faster services to be consumed and removed.
The bit that really did catch my eye though was the fact that storage is not the traditional storage array, when I was installing those FlexPod systems 5 years ago they would consist of a FAS controller plus disk shelves, a couple of Cisco Fabric Interconnects, Nexus Switches and UCS blade chassis. With the FlexPod SF offering it takes consists of “Storage Nodes” these storage nodes are Cisco C220 M4 servers containing some SSD, 10GbE interfaces.
SF9608 Storage Node
- Built-on Cisco C220 M4 Chassis
- 8 Samsung 960GB SSDs (non-SED)
- 10GE iSCSI only solution
- Supported within FlexPod Configuration only
- Minimum 4-node cluster
Hold Up. A C220 is 2U, minimum 4-node cluster means 4 x 2 which is 8U to begin with. There’s your maths lesson today kids. But wait theres more.
NetApp and Cisco are saying that these 4 nodes, 8U configuration will provide you with 300,000 IOPS and 30TB effective storage. That’s impressive.
But then you are also able to add single nodes to your cluster effectively giving you 10TB and 75,000 IOPS with each increment.
From a compute point of view, it uses the same Cisco UCS B-Series M4 servers found in the traditional FlexPod as well as being able to connect to either the UCS 6300 or the 6200 Cisco Fabric Interconnects.
Converged Infrastructure – A Validated Physical Topology
As I previously stated the components are shown below, as with all converged infrastructures they should be validated and have a reference architecture to validate its capabilities and that it serves its purpose.
One thing for sure is whilst out at Cisco Live this week I will be catching up with the SolidFire guys to understand more about this. I am a big advocate of the FlexPod storage from both Cisco and NetApp and this is only going to add more options and capabilities to the FlexPod Portfolio.