January 2011

“Storage Jumping” with EMC

January 18, 2011

Peter Eicher

Post by

Senior Product Specialist, Data Protection

On a cold, soggy day in New York City, EMC rolled out a series of “record breaking” announcements. Lots of stuff, most of which I’ll leave to others to comment on.  A lot around “big data” and things like that, and certainly nobody disputes EMC’s status in big iron.

But one interesting item was the VNXe, a new line of unified storage for small business.  As seems to be the way with these sort of huge, corporate blowout events, the VNXe came along with a silly video featuring a 5th grader claiming to be able to provision the device. Because every data center I know is crammed with 5th graders!  And then to make it worse, they dragged the kid on stage with an iPad so he could mumble a few remarks about managing the product, convincing no one that he had the slightest idea what he was doing.  He should go back to playing Angry Birds.  I won’t even discuss when they crammed 26 people into a Mini Cooper to prove… that you could cram 26 people into a Mini Cooper.

In any event, the one sentence VNXe tagline from the EMC website has an eerily familiar ring to it. 

Unified storage platforms with one management framework supporting file, block, and object – optimized for virtual applications.

I dunno, where have I heard that before?  Maybe… NetApp?  Yup, this sure seems like an attempt to get into the space that NetApp has owned for years.  Glad to see EMC admitting the world needs unified storage. And now EMC is truly “unifed”… across one product line.  Maybe next year.

The other interest of mine was the new Data Domain DD890, a massive pile of iron for deduplication.  EMC is touting the “world record” of 26.3 TB/hr data ingest rate. Impressive, but limited to a pretty rarefied group of users.  The 10 Gb interfaces alone will limit the user base. There are other problems as well.  How do you get your servers to push data at this rate?

The thing about a 26 TB/hr ingest rate is, you’re asking your servers to lift 26 TB an hour of data, and your network to carry it. Why would you want to do that?  A big dedupe canister at the end of your backup cycle is all well and good, but it does nothing to limit backup impact at the host or on the network.

This is why at Syncsort we say you need to go beyond deduplication to solve the full range of data protection and recovery challenges. Dedupe is great and necessary, but as the logicians say, it is necessary but not sufficient.   

Finally, my favorite part of the EMC event was yet another “record,” when motorcycle daredevil Bubba Blackwell jumped his Harley over 40 EMC VMAX storage systems.

As Dave Barry would say, I’m not making this up. Apparently, this is the new world record for storage jumping. Really, what can you even say?  I prefer the other kind of storage jumping, the one where EMC customers jump to Syncsort and NetApp for a better data protection solution. And lately, that’s been happening a lot.

{ 0 comments }

Simple, Integrated and Proven Data Protection…What’s Not to Like!

January 18, 2011

Peter Eicher

Post by

Senior Product Specialist, Data Protection

The always interesting Steve Duplessie of Enterprise Strategy Group (ESG) posted recently about some recent ESG research in his blog, The Bigger Truth.  He covers a lot of territory since his turf is the entire IT pie. My turf is only the small slice of data protection (if you consider a $6 billion annual spend “small”!) so I’ll limit my comments accordingly.

Steve pointed out a couple of findings from ESG’s latest research:

  • 58% of organizations have virtualized less than 1/3 of their servers  
  • 59% haven’t virtualized ANY “mission-critical” applications

We all know virtualization is the biggest trend in IT, but it’s also clear there’s a long way to go.  Many users are still struggling with various aspects of their virtualization roll out, and they are at risk of not realizing all the potential value. Duplessie writes:

“We can put all the VMware in the world all over our environment and without the ability to AUTONOMOUSLY manage it, it will always suck.  Why? Because it will always come down to the lowest common denominator–the ultimate point of failure–a person’s ability to make the right decision at the exact right time and execute it perfectly.  It’s unreasonable.”

Right enough. From my data protection perspective, it’s clear that a “set it and forget it” data protection scheme is critical for virtualization. Backup has long been a giant black hole when it comes to IT work hours because of the tendency for backups to fail (and fail and fail) or run too long. Much of that has to do with the multi-hop nature of legacy backup. The agent sends data to the media server which sends it to the tape library or disk, with various switch layers in-between. When something goes wrong, how do you hunt down the source of the problem?  Trouble spots are often in the connections between hops, because it always gets tricky when you need to connect disparate parts from different vendors. Problems can come down to something as mundane as the wrong HBA firmware in your media server (your disk target was never certified on that version of firmware… who knew!).  

The virtual world has only added to the complexity, because now you’ve got virtual switches on top of physical switches, virtual operating systems on top of hypervisors, and so on. It’s like a hall of mirrors. The fact that some virtual machine backup solutions add proxy servers into the mix only aggravates the situation.

At Syncsort, we like to keep it as simple as possible. One of the key design advantages of NetApp Syncsort Integrated Backup (NSB) is that it eliminates the entire media server (or device server) layer. It’s just not in the architecture.  Syncsort agent software sends backup blocks directly to the NetApp FAS storage systems used as backup targets. No proxies, nothing else in the data path, just a straight shot to the disk.  When NSB creates snapshots to store the block level backups, we don’t use a server running backup software to drive the actions of the disk system. We use the native, rock-solid NetApp Snapshot code (direct integration with the firmware). When we mount a snapshot for instant data recovery, we don’t run some kludgy code off a server that needs to work with a hundred different disk arrays. We invoke the NetApp FlexClone technology. Super high performance, proven in thousands of customers every day.

This is what we mean when we say that NSB is simple, integrated and proven.

Simple: removing as much complexity as we can, limiting the number of moving parts. Easy to deploy, easy to use.

Integrated: leveraging the best technologies that Syncsort and NetApp have to offer. Why would Syncsort want to write new snapshot code when NetApp’s is the best in the industry? Simple answer: we don’t. We didn’t. We do what we do best, and we make use of NetApp’s best-of-breed disk technology and features.

Proven: It’s out there, protecting customers every day across every industry.   

The bottom line is a level of backup success that sometimes shocks our new users. 99% and better backup completion rates are common. What used to be hours and hours of weekly troubleshooting drudgery simply disappears, letting the IT folks worry about delivering value added services and applications to their organizations, not wondering if a firmware upgrade is going to solve a mysterious failure or make it even worse.

And all this is that much more important as you move deeper into virtualization. Minimize the complexity.  Deploy something that just works. Stop trying to figure out if what you’re looking at is the real problem or just a confusing reflection.   

Sometimes the best way out of a hall of mirrors is with a hammer.

{ 0 comments }

Today’s Forecast: More Data, Less Budget

January 5, 2011

Jorge A. Lopez

Post by

Senior Manager, Data Integration

Big data for small budgets: that seems to be the outlook for 2011. In fact, this is a trend we’ve seen for a few years now. The recession has struck every department within organizations, and IT has not been an exception. While IT spending remained relatively strong in 2010, most experts are predicting that budgets will probably grow little – if at all – in 2011.

At the same time, according to a Gartner survey cited in a recent article from Laura Smith (Budgets, big data, and staffing among top CIO concerns) data volumes are expected to grow more than 800% within the next five years. As we mentioned in previous posts, this growth has been largely fueled in recent years by 2 factors: mobile technologies and a tremendous increase of content producers thanks to new collaboration paradigms.

So, where will most of the IT dollars go?  Unfortunately, a big chunk of our budgets are still going to hardware and staffing to maintain SLAs on existing initiatives. However, as organizations are challenged to do more with less, they will turn to solutions that can scale to support an exponential data growth without corresponding increases in staff, hardware, and software. In other words, scalability is not only about performance, it is also about costs.

This is why increasingly more organizations are turning to Data Integration Acceleration solutions, such as DMExpress. Syncsort’s definition of scalability means:

  • Zero tuning required
    • Dynamically adapt to changing conditions such as number and type of sources
    • Automatically select the most efficient algorithms for your data
  • No massive hardware investments
    • Fastest performance on commodity hardware with minimum CPU and memory utilization
    • Automatically optimize performance based on hardware environment

Yes, zero tuning required! How many hours does your IT staff spend tuning and re-tuning your DI environment? How much additional hardware are you planning to buy in the next year to maintain your SLAs? How are increased tuning and increased hardware helping you reduce costs? As CIOs grow concerned about costs and IT staff, it might not be too late to rethink the strategy.

{ 0 comments }