Why Backup Is Big

Within the storage-related disciplines, the folks responsible for data protection -- backup, replication, recovery, etc. -- historically haven't gotten their fair share of the credit they deserve.

The perception that data protection was important, but certainly not glamorous.  Sort of like the folks who do tax audits -- important, but not glamorous.

A number of trends seemed to have reversed the pecking order.  All of the sudden, backup is hot.

Why Am I Saying This?

I'm seeing evidence everywhere. 

For example, I get to see just about every IT survey that's done in the industry.  And, yes, data protection projects have now risen to one of the top storage-related projects, and even are starting to break in to the top 10 for the broader category of "IT priorities".

Further evidence comes from my discussions with customers and partners: data protection projects are very popular right now, and don't show any sign of abatement anytime soon.

Many of the service providers I work with also point to strong demand for data-protection as a service -- one of the few areas where there's a large market and many clients who are more than happy to give this task to an external specialist.

And, finally, there's the activity level within EMC's own BRS division (backup and recovery solutions) which continues at a frantic pace -- not to mention the adjacent replication and archiving products.

Why Is This?

When something gets unusually popular in our industry, I always want to understand "why".  Just my inquisitive nature, I guess.

I can point to at least five secular trends that are working in concert here.

First, we're in the middle of a technology transition from tape to disk.  The magic is software that deduplicates data, and makes disk the far more attractive option for far more people than before.

One of the industry studies I get shows the aggregate market for tape libraries and tape media year over year.  Every study shows mostly the same thing: the market for both shrinks anywhere from 20% to 30% or more -- depending who's survey you're reading.

Most tape libraries out there are getting old, and either the lease is up, expensive maintenance is beginning, or both.  And more and more IT professionals can't bear the thought of buying more tape technology when better disk-based solutions have been out there for a while.

Second, ongoing backup costs are easy to target from an expenditure reduction perspective.  It's pretty easy to get to capex/opex expenses associated with the activity, and intelligently evaluate alternatives.

Third, data volumes continue to grow, and grow, and grow.  No one seems to be escaping this inevitable trend.  Sooner or later, the backup/replication solution that was designed three or five or seven years ago just isn't keeping up, architecturally speaking.

Fourth, there's a lot of re-platforming going on these days -- primarily from physical to virtualized environments.  As part of the process, there's usually a revisiting of all the core infrastructure elements -- server, storage, network, etc. -- and data protection is part of this discussion as well.

Finally, protecting data is one of IT's core responsibilities. 

Imagine a CFO who said "well, we lost a bunch of money, we're not sure what happened, and we're working on getting back most of it".  Wouldn't be good for the CFO's career, would it?

Same thing happens when senior IT leadership has to turn around and say "sorry, we lost a bunch of your important data, we're not sure what happened, and we're working on getting back most of it". 

Not a pleasant discussion for anyone to have.

All of this has made the entire topic of next-gen data protection very popular, and the IT pros who specialize in these disciplines very popular indeed.

Service Catalogs Rule

Some IT organizations are lucky enough to have modest and/or homogenous data protection requirements, and are definitely looking for a one-size-fits-all approach.

If you're lucky enough to be in this situation, there are good data protection products (from EMC and others) that can cover a reasonable spectrum of IT requirements using a single product approach.

However, in larger organizations, that tends to be the exception rather than the rule.  They have to think about the problem differently -- they need to think in terms of a data protection service catalog that balances RTO/RPO and costs.

In case the lingo escapes you, RPO refers to "recovery point objective", e.g. how much data can you afford to lose because it's not protected?  There is no right answer -- shorter RPOs inevitably drive up costs, with a steep incline in costs when the answer becomes "no data can be lost".

RTO refers to "recovery time objective" e.g. how long can you wait to get your data back and get back to business?  Again, no right answer, things get more expensive as required times decrease, and -- again -- a steep incline in costs when the answer becomes "no downtime is acceptable".

Indeed, the skills and methodologies associated with understanding business requirements and coming up with a reasonable data protection services catalog is a healthy industry unto itself these days ...

Sometimes you can generate a decent services catalog from a single data protection product, like Avamar and/or Recoverpoint.  More frequently, though, multiple technologies are indicated, and ideally orchestrated with something like EMC's Data Protection Advisor.

I wish the world was as simple as some single-product vendors make it out to be, but that's just not the case for most of the IT landscape.

Technology And Process Redesign

If the goal is to deliver "data protection as a service" (which it should be), it's not enough to simply buy a bunch of the new kit and call it a day.

To get the best results -- especially in larger environments -- we're talking about some non-trivial process redesign in support of the new service catalog: provisioning, monitoring, validating and costing. 

It's not a complicated engagement as far as such things go, but not making the investment usually results in less-than-optimal results.

It's Great To Be EMC These Days ...

I don't know whether we were smart, lucky -- or a little bit of both -- but right now we seem to have the strongest and most complete offering in this space. 

From where we (and many of our customers) sit, there's currently a very wide gap indeed between what we're doing, and what everyone else is doing.

It starts with great products: DataDomain and Avamar are class-leaders in their respective categories: target and source dedupe.  For replication-oriented requirements, Recoverpoint delivers a wide range of continuous replication capabilities using a simple, industrial-strength and storage-agnostic appliance.

Add in the wide selection of traditional data protection technologies: Networker, SRDF, Timefinder, Mirrorview, Celerra Replicator, HomeBase, etc. -- we've got a *lot* of tools in our toolbox -- far more than any other vendor.

And then we start bringing out the secret sauce.

Part 1 of the secret sauce is EMC's popular Data Protection Advisor -- think of this as a orchestration layer for delivering data protection services regardless of the underlying mechanisms being used -- even if they're not from EMC.

Nothing like being able to look at a single screen (or have your clients look at single screen) and be able to quickly say "yep, everything is protected as intended".  I, for one, think this leads to more restful sleep by the IT organization -- and their users ...

Part 2 of the secret sauce is a complete portfolio of consulting, implementation and management services. 

We have people who can get you to your requirements quickly, since they do it every day.  They can even help you discover application and data dependencies behind important business processes -- something that's notoriously difficult to do.

We have people who can help you evaluate different alternatives and scenarios using easy-to-get-to metrics.  We have people who can quickly implement the new solutions, migrate your workloads, and show you how to do it.

And, if you'd like to consume data protection as a managed service (your facility, our people), well, we're doing a lot of that as well.

No wonder EMC's business is doing so well in this area -- there doesn't seem to be any comparable vendor offerings that are so advanced, so complete and so implementable. 

Fortunately, we're able to offer our customers incredibly complete customized and turnkey solutions -- and not just one or two point products here or there.

Recent News

Many in our industry saw our recent acquisition of Bus-Tech as intriguing.  Other folks might wonder why we'd invest in mainframe-oriented data protection solutions.

The reasons are simple if you think about it. 

First, there are a lot of mainframes out there, and the data on them tends to be rather important stuff. 

Second, these are mature environments, and rarely does the opportunity come along to take some aspect of the mainframe environment and do it seriously better -- as is the case here with data protection. 

Third, customers tend to want complete solutions to data protection, and not just piece parts.  If part of their landscape includes mainframes, they want products and solutions that cover the entire portfolio, and not just the convenient-for-the-vendor parts.

EMC has worked with Bus-Tech for many years.  My experience has always been that they were great guys with a great product.  As part of the EMC family, there's a clear opportunity to do more of what they do so well -- and that's a good thing.

The Bottom Line

If information is getting more important (and it is), then protecting it is going to get more important as well. 

The newer technologies have already proven their transformational power to deliver data protection services that are significantly faster, better and cheaper.

And, yes, now would be a good time to thank your data protection team. 

They're the only thing that stands between you and a really bad day.

Comments

[P. Clarke Thomas]

I don't believe tape will ever disappear. Depending on the organizations historical requirements, e.g., HIPAA, SEC, etc, would help to determine how much data is kept at quick/easy access(disk) vs tape. Plus if you were a government agency or needed to keep ancient data, you wouldn't keep 20+ year old data on disk if it's rarely being accessed.

That being said, I did love the Centerra I was last using for dedupe & archival of low access data. As for live data, for smaller organizations I can see the need for basic backup. Are we possibly, now at the point where the consumers(users/mgmt) want more instantaneous access....such as purely switching to a new datacenter (within minutes) with little to no data loss. Thus no longer rebuilding machines, more so just powering them on.

[Rob]

"One of the industry studies I get shows the aggregate market for tape libraries and tape media year over year. Every study shows mostly the same thing: the market for both shrinks anywhere from 20% to 30%"

I wonder how much of this is due to LTO capacities doubling
every few years?

But I know of largish (for us) customers that are forced to do
blended de-dupe solutions. Large databases and large amounts
of images don't lend themselves to de-dupe. Also, disk
isn't as green as LTO. I've read elsewhere that LTO tape is
230 times as energy efficient as disk, and 3 times more cost
effective (not sure about exact numbers, I suppose I could
google it). Also, SOX compliance explodes retention times
and a number of de-dupe long term retention work-arounds
are brutal.

Finally, LTO5 native capacity is 1.6 TB, at least 3 TB
compressed. 3.2 TB native for LTO6 in the future. I'd think
LTO makes sense for most customers. Greener, high capacity
and pretty cheap in comparison especially for long-term
retention requirements.

[visiotech]

You might listen latest InfoSmack on Storage Monkeys about Tape vs Disks and my comments at the bottom too.

In my view we might see high performance spinning disks disappear faster than tape. SSD will take a much bigger place in less than 5 years.

I see good marketing strategies for those who offer only disks solutions and do not manufacture tape.

But tape manufacture are much less today than it used to be. So their offering is must less and R&D too. They did not fix most media handling and expand on dedup. Bottom line they could have done a better job understanding what are customer pain with physical tape.

Virtual tape and other technologies are great for 24 to 72 hours. Somehow tape fit very well by protecting data against just about any threat can happen in the data center when encryption is turned on. Tape protect data very well against several admin errors, virus, fire, water-leek etc Not always the case with disks.

I love backup to disk because most servers cannot stream LTO-4 or LTO-5 tape at half of their native speed when millions of files are on your disks. Strangely no benchmark are done since LTO-3. They all compare VTL or disk solutions with old stuff.

Long live to tape...and hope SSD replace spinning disk.

[W. Curtis Preston]

@Chuck

I've been telling me kids that backup is sexy for a long time. They agree cause it pays the mortgage. ;)

Having said that, I don't know how much I agree with that annual 20/30% decrease of tape. If that were the case, the market would disappear in under 3 years.

It's obvious that more and more backups go to disk, but the bulk of backups are ALSO still going to tape. VERY FEW people are going tapeless, if for no other reason than it's much more expensive than a combo tape/disk system.

@Rob

Not sure where you heard that large databases don't dedupe well. They dedupe great; if they don't your dedupe solution is crap. Images/video/audio, etc. don't dedupe, though.

You read that power stuff in the clipper group report. I debunked that report (or at least much of it) on my blog when it came out.

But LTO does make a lot of sense for a lot of use cases. You definitely got that right.