Q&A on Flash with Brocade

At this week’s Flash Memory Summit, we will be talking extensively about our comprehensive portfolio of hybrid and all-flash storage solutions.  In addition to the relevance of flash to application acceleration and data center efficiency, we are often asked about the networking considerations for this technology space. With this in mind, we thought it would be helpful to discuss this topic with Rick Balderrama, Director of Partner Marketing, from our networking partner Brocade.


NetApp:  When customers are thinking about deploying Flash storage, how should they think about the network?


Rick (Brocade):  We’re finding that customers are incredibly busy just trying to keep up with the changes in their environments.  They aren’t able to spend a lot of time thinking strategically about Flash and the infrastructure to best support that storage.  Implementing Flash isn’t about the network, or even the storage, it’s about the application and customer experience.


NetApp:  That’s correct, it’s all about the application.  Customers are implementing NetApp Flash because of the performance improvements.  So, does the network make a difference?


Rick (Brocade):  Absolutely.  The challenge customers have to overcome is often due to internal responsibilities.  The storage folks are thinking about the storage, and often don’t have time to think about the complete application environment..  In data center environments you’ve often got storage admins implementing Flash, you’ve got network admins doing their thing with the network, you’ve got the server guys assigning VM’s to apps, and the application folks trying to get the most out of their application investment.  The intent of implementing Flash is for improved experience of the internal or external customer when using the application – but the entire infrastructure has to be tuned correctly to get the most out of that Flash storage investment.  The silos of responsibilities in many data centers don’t always lead to making the right choices on the network to maximize the benefit of the storage to the application.


NetApp:  So, is it a matter of tuning an existing network infrastructure, or is it more than that?


Rick (Brocade):  One of the significant changes in the data center is the focus on IOPs.  This is really driven by virtualization.  So, while storage folks have historically been focused on bandwidth, for example 16 Gbps Fibre Channel or 10 GbE, the focus is now also including IOPs.  

A good metaphor our engineers use when discussing IOPs versus bandwidth is a highway.  The bandwidth is like the highway.  The more Gigabytes per second, the more lanes.  IOPS would then be the number of cars that can run at speed in those lanes.  The closer they can run together at a faster speed, the more IOPS you get.  So, with Flash you need to consider IOPs not only on the storage, but also in the network to ensure you are getting the performance you want for your application.

NetApp:  Is it just IOPs and performance, or are there other considerations.

Rick (Brocade):  Availability is also really important – again because of the application.  The majority of applications that are being implemented on Flash are mission critical apps.  So, performance is important, but performance without availability doesn’t support an application that’s mission critical.  A lot of customers get excited about the performance, which they should be, but they can lose sight as to why they are implementing Flash in the first place.  It’s to support a mission-critical app.  So, you need availability too.

NetApp:  So, we need a network with high performance and high availability.  What are thoughts on network protocols then?

Rick (Brocade):  Brocade provides traditional Ethernet, Ethernet Fabric and Fibre Channel solutions to customers through NetApp, so we’re agnostic on the protocol.  But we’re finding the best solution for Flash is Fibre Channel, and specifically Gen5 Fibre Channel which provides 16 Gbps bandwidth and up to 4M IOPs. 

In fact, Gen5 Fibre Channel is used by VMware in 100% of its top 20 benchmarks, and 83% of VMware deployments are on Fibre Channel.  Additionally, it’s got the 5 9’s availability that’s been proven in data centers supporting mission critical apps in the largest customers in the world.  On top of that, NetApp provides Automated Support (ASUP) which is an integrated call-home feature for the storage and the Brocade Fibre Channel network.  This means the data center can have policies set so the storage and fabric that will automatically call NetApp support for help, and potentially have an issue resolved before a human was aware an issue existed in the first place. So, the combination of the 99.999% availability, the high IOPs, and the enterprise level of support really makes it the right solution for those mission critical apps running on Flash storage.