Blogs

The Real Definition of Future Proofing

By Mark Adams posted 10-09-2019 07:00

  
One of the challenges that data center executives struggle with is how to transition from what works today to embracing emerging technologies that hold great promise in their ability to address the ever-more-stringent SLA requirements that are coming from their business partners.  Past examples of this include the transition from direct attached to SAN attached storage.  The move from physical servers dedicated to hosting a single application over to virtualized servers that consolidate multiple applications is another.  Unfortunately, the landscape is also littered with many unsuccessful technology transitions that have been proposed over the years but failed to deliver, FCoE being just one example.  In all of these transition opportunities, data center executives are faced with answering key questions: a) how much benefit will new technologies give us over what we’re using today, b) is the new technology viable and ready to take on our workloads without introducing any unacceptable levels of risk and c) what are the tradeoff such as additional costs or loss of functionality that will detract from those promised benefits?  

In today’s storage environment, NVMe has been developed as a new interface for transmitting IOs between the storage controllers and flash media, what storage folks refer to as the “back-end” interface.  It’s been designed to replace the SAS interface which has been in use for more than 10 years now.  A related technology is NVMe over fabric which brings that same protocol stack to iSCSI and fibre channel storage networks which storage folks refer to as “front-end”.  The third technology that’s coming is called Storage Class Memory (SCM) which is a new type of non-volatile flash media that promises a significant reduction in latency as compared to today’s NAND-based SSDs.

Each of these three technologies was designed to improve the speed of all-flash arrays.  For applications that demand the highest level of performance, data center executives would be advised to evaluate their potential.  And each comes with its own set of trade-offs.  Back-end NVMe will add some cost to the price of the storage array, has lower density until NVMe-based SSDs catch up to SAS-based SSDs in capacity and is less scalable because NVMe disk trays cannot be daisy-chained together like SAS disk trays can.  Front-end NVMe requires that the array be connected to a storage network that supports an NVMe fabric.  The Brocade Gen 6 switch is one of the few options available that supports NVMe over fabric.  Without that support in place, you’ll need to upgrade your fabric as well as your array to take advantage of NVMe over fabric.  And SCM costs many times more than today’s NAND-based SSDs so only the most performance hungry applications that can realize a significant business return for this investment should consider SCM until the pricing erodes.  

However, what if you could mix and match between today’s battle-tested technologies and these promising future technologies?  What if you could be confident that you can keep your day-to-day operations running non-stop on traditional technologies while using emerging technologies to store newly developed applications that rely on the fastest analytics possible to keep you ahead of your competition and ensure the highest levels of customer satisfaction.  What if you could do all of this in the same system and manage it as a single entity?



This is exactly how we designed the VSP 5000 series.  These new enterprise systems, VSP 5100 and VSP 5500, from Hitachi were designed as a node-based scale-out architecture with a high speed and hardware accelerated fabric that connects the nodes together.  The nodes are configurable in pairs and we support heterogeneous interfaces across node pairs.  For example, the first node pair can be ordered with a traditional fibre channel front end and a SAS backend while the second node pair can have an NVMe over FC front end and NVMe backend.  Any node pair with an NVMe backend can have SCM as well as SSD media intermixed within the drive trays.  Not all of these options are available today as SCM and NVM over fabric will be coming next year.  However, the VSP 5000 series that we ship on day 1 will be seamlessly upgradeable to these new technologies when they are ready to be released.  SCM will be compatible with the NVMe drive trays and NVMe over fabric will only require an update to the fibre-channel port drivers.

Which other enterprise storage systems can provide you this level of flexibility to take you from what works today to what’s needed for tomorrow’s storage requirements and do so in the same system?  This is why Hitachi is so excited to be launching the VSP 5000 series as we see it as a complete game-changer.  And the customers who tested the early eval units have been wowed as well.


#ThoughtLeadership
#HitachiVirtualStoragePlatformVSP
#Blog
0 comments
1 view

Permalink