Clearpath’s Blog on IT Infrastructure, Hybrid Clouds and IT Security

At Clearpath, we work with a number of community hospitals.  Like all verticals, healthcare has its own set of unique elements and challenges - the most important of which is related to system downtime.  It’s not a revelation that downtime costs money; that is well-documented.  However, in a hospital setting, downtime has the potential to impact patient health and well-being.  Imagine the impact of downtime when trying to register a new patient in the ER, when pulling up an MRI in a patient room, or requesting medication from the on-site pharmacy.  All of these tasks depend on reliable backend IT systems.  Although system downtime is a component of any IT environment, it is especially important for hospitals to invest in business continuity strategies that will minimize the amount of planned and unplanned downtime wherever possible.   

EMC has announced the latest software update to its Data Domain platform – DDOS 5.3.   It seems like there is something for everyone in this upgrade, so make sure to review the new functionality to see if there is something there for your environment.  I’ve listed out some of the highlights below.

RecoverPoint is EMC’s flagship replication product.  It is a purpose-built combination of appliances and software that replicate block data between arrays.  It boasts WAN-optimization and the ability to roll back and forward in time (think DVR) to present the replicated LUN as of almost the exact time that you wanted (invaluable in the case of corrupted data).  It is a popular replication mechanism for users of VMware Site Recovery Manager.  Fully integrated as a Storage Replication Adapter, RecoverPoint replicas can be controlled and mounted via SRM.

EMC VPLEX is a one of a kind solution that provides site resiliency and high availability for those that can’t afford any downtime.  VPLEX itself installs in the network, in between hosts and storage, and can extend data over distance, within and between data centers.  EMC refers to this technology as Federated Access Anywhere.

In the ‘Metro’ version, VPLEX front-ends an existing array (EMC/non-EMC) at two locations and provides the ability for both to have an active copy of a single set of data.  This active-active model is facilitated by a something called cache coherency and provides a distributed virtual volume to which servers at both sites can connect.  Stretched clusters anyone?

While both RecoverPoint and VPLEX are great technologies with specific advantages, storage admins have had to choose which one they wanted on a per LUN basis: For this dataset, do I want the high resiliency of VPLEX that allows me to vMotion over distance in the event of a site issue? Or, is the ability to roll that data back to a point in time more important?  Up until now, it has been one or the other because these two technologies didn’t easy fit together.  Thankfully that is now changing with the latest software versions of both VPLEX and RecoverPoint.

The combination of RecoverPoint 3.5 and GeoSynchrony 5.1 (VPLEX’s OS) now allows you to replicate a VPLEX volume the same way that we have been using RecoverPoint to replicate a LUN from a standalone array.  The difference is the inclusion of a RecoverPoint write-splitter inside VPLEX - something that had been missing up until now.

Hard drive sizes have increased again.  It is just part of the evolution of technology.  When it comes to using them in storage arrays however, bigger is not always better – or at least may not be the best course of action.

This morning I attended Pat Gelsinger’s Keynote at EMC World. Definitely like his presentations as he’s always heavy on content and examples but light on fluff. At the end of all of the iterative product announcements (which are all meaningful in their own right), he outlined something in the works called “Project Lightning”. Chad Sakac from EMC does a technical dive of it here. The gist of it is this – there are specific performance use cases where we want to move the data closer to the compute resources (less latency) or the compute closer to the data (higher bandwidth).

Many of our customers initially purchased single-shelf Data Domain DD630 appliances with 7 drives, with the intent to upgrade them later to 12 drives when additional capacity was required. Once the drive upgrade has been ordered and shipped, the default process to utilize the additional capacity is pretty straightforward. The only problem with the default capacity upgrade method is that 2TB of additional space is sacrificed to parity. In essence you would be adding only the capacity of 3 drives out of the 5 you purchased. Unlike a Clariion or a VNX, the Data Domain does not have the ability to expand its disk group. Instead, it creates a second RAID 6 disk group and just extends the overlying file system. Again, the downside here is that an additional 2 drives are being allocated to parity when a second disk group is created.

I have never seen the buzz surrounding a product announcement like I did with the launch of EMC's new storage products on January 18th. Don't get me wrong, I got caught up in it as much as anyone else – glued to the webcast as EMC intermingled new product introductions with world record breaking stunts (BTW - You could not have paid me to be one of the 26 people stuffed into a Mini Cooper like folding chairs). That being said, what does the release of VNX mean to your storage environment? You'll have to decide for yourself but here are the top five things that I am excited about:

1. With vSphere 4.1, VMware has been kind enough to move vMotion down into the Standard and Essentials Plus software tiers. This is a real bonus for shops running these versions as they now have access to the key component that enables the delivery of higher uptime.