What Does a Transformed Backup Environment Look Like?

As I’ve roamed the halls of EMC World over the past two days, the most frequent question I’ve been asked is “What does a transformed backup environment look like?” (Actually, more people have asked me what Zolla and I were doing on stage during the keynote? So, for the last time, his mike wasn’t working and I was letting him use mine to tell the people in the booth, Ok?)

Most people are expecting a blueprint for an answer. I imagine they expect me to pull out a Visio diagram teeming with DPA, NetWorker, Avamar, DLm, and Data Domain. But backup transformation isn’t defined by any specific technology or workflow. You don’t achieve “transformation” status by deploying Oracle RMAN backups directly to Data Domain, leveraging VMware Changed Block Tracking for Avamar backups or even by going tapeless. You do it by:

  • Establishing trust between the backup team and its consumers. The VM, application, primary storage, and server teams feel that they are partnering with their backup team. There is communication, innovation and understanding between them.
  • Addressing “safety net” concerns and accelerating business velocity:  Your organization has such confidence in its “safety net” (i.e., its backups) that data protection concerns aren’t slowing down key initiatives (e.g., virtualization, application consolidation/refresh, storage consolidation). Instead, your organization is moving even more quickly because it knows that the safety net is there to protect it.
  • Thinking of backup as a business-enabler: Instead of worrying about being marginalized by other teams, your backup team knows that it is adding business value. Instead of spending time trying to keep its head above water, your backup team is making strategic plans for the future (e.g., evaluating cutting-edge backup approaches, exploring archival strategies and discussing other uses for the protection copies of their data).

But, most importantly, you know you’re on track, when your backup team embraces a service-provider-style model. Instead of insisting that backups and recoveries run through their heavyweight processes and infrastructure, the backup team delivers appropriate solutions to meet customer challenges and corporate protection standards. This may mean enabling an Oracle DBA to run its own backups/recoveries via RMAN to Data Domain, protecting 10 other databases with NetWorker and DD BOOST, and running DPA reports to manage the entire environment.

Of course, as with any big change, taking the first step is often the hardest.  But trust me the journey is well worth it!

Stephen Manley

Stephen Manley

CTO, Data Protection and Availability Division
Over the past 15 years at both EMC and NetApp, I have traveled the world, helping solve backup and recovery challenges - one customer at a time (clearly, I need to optimize my travel arrangements!). My professional mission is to transform data protection so that it accelerates customers’ businesses. I have a passion for helping engineers pursue technical career path(without becoming managers), telling stories about life on the road and NDMP (yes, that’s NDMP).

One thought on “What Does a Transformed Backup Environment Look Like?

  1. You’re spot on. Best practice for ruinnng Oracle Database virtualized is to run it in a dedicated vSphere cluster for just the Oracle Database. That way you get the best bang for the buck. Since all the cores need to be licensed to run the Oracle database, why would you want to also put exchange or sharepoint on the same cores? The beauty of Oracle’s licensing is that once those cores are licensed you can run as much Oracle as you can fit on them. Therefore, having modern day processors with a lot of memory is a great way to consolidate Oracle. The vBlock is all based on the latest Intel Westmere Processors and the Cisco UCS servers in the vBlock you can grow a single dual core blade to 384 Gig of RAM, which is pretty impressive for a blade. The Vblock is designed to be highly partionable both at the physical level creating separate physical clusters then at the logical level using VMware vSphere and vCloud Director for further partitioning of the resources. You’d perform physical partitioning to manage your licensing to Oracle, then virtual partitioning to provide multi-tenancy across companies or across LOB’s or across Dev, Test Production, etc… The choices are yours to do what makes the most sense.As for support I do I have a few thoughts on that. Oracle does in fact support the Oracle database ruinnng on VMware, and they’ve even recently altered their support statement stating that they’ll also support RAC ruinnng on VMware (see Oracle’s support note: 249212.1). What scares most customers is that they also say that if they don’t think it’s an Oracle problem they may ask the customer to reproduce the problem on a physical server. This is not so different then what they could do on a lot of platforms. If for example you’re ruinnng Oracle on Windows and hit an issue that Oracle believes is a Microsoft bug they could push back and tell you to go to MS. It’s kind of the same thing. They’re either saying, “show us it’s not VMware by showing us it happens without VMware, or go call VMware because we Oracle think it’s a VMware issue”. In those cases, having to call VMware isn’t necessarily a bad thing. You probably want to call VMware in parallel with calling Oracle anyway if a tier 1 production workload is having a problem. VMware has an Oracle support statement that (if I can paraphrase) VMware will own your support case and work directly with Oracle to get it resolved. If it is a VMware issue they’ll fix it and if it’s an Oracle issue they’ll help Oracle see the way. In my 10+ years using Oracle on VMware I’ve only heard of a few times (I can count them on my fingers) Oracle threatened to have an issue reproduced, and most of those situations were resolved with a little push back. Oracle also seems to be getting more lenient requiring less situations like this than a few years ago. I know of one large fortune 100 company that got an Oracle Executive to put in writing that they would fully support VMware.If this still isn’t reasonable to any remaining users out there, there are many published best practices for easily swinging an Oracle database from virtual to physical. Probably the most common way is to use raw device maps (RDM’s) to point a physical LUN to a VM, then take that same physical LUN and point it to a physical server. What less people know about is that with Oracle 11g, Oracle introduced Direct NFS (dNFS). dNFS is not quite as fast a block, but a lot better than traditional NFS. Modern hardware such as the vBlock support 10 Gig Ethernet, so ruinnng NFS over 10 GbE can be a nice alternative to block, and it’s actually easier to manage and maintain because after all it’s just NFS. With NFS any server, physical or virtual can access the files, so you can have your Oracle executables, config files and the data files themselves all in NFS, and start that DB with either physical or virtual resources. The vBlock allows you to partition blades as either physical blades or virtual blades, so you can keep everything contained in the vBlock and have a spare server used for this purpose. Furthermore, you can take a snapshot of the database and boot the snapshot with a physical without incurring a penalty on doubling the size of the database because a snapshot is just the delta changes. Then Oracle get’s what they want, the customer gets the timely support they want, vBlock supports it all, and everyone wins.I hope that gives you some things to think about, but please feel free to comment if you have any further questions.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>