Bruce Backa's thoughts on technology and life - Storage Management

Another Technology Passes

Your DisplayName here! — Mon, 03 May 2010 15:47:13 GMT

Depressingly (for me, anyway),today's generation doesn't even know what floppy drives are. In another 10 years the same will be true of 'disk' drives or 'hard' drives. This year is the beginning of the end for rotating media. You can already go to Staples and, for a reasonable price, buy an 80-gig solid state drive that is 100 times faster rotating media. Once there are no more rotating drives in desktop computers, servers will follow in a couple of years. Goodbye hard drive... you've served us well.

Cloud? What cloud?

Your DisplayName here! — Thu, 12 Nov 2009 14:43:06 GMT

The bad news is that for this post I have to admit to being over 40 (maybe even over 50…).

These days I’m often fascinated to see what 30-something MBAs in high tech marketing can come up with – and how much money they get to spend once they do. The latest, of course, is Cloud Computing and Cloud Storage. What is Cloud Computing? All your data and your applications are somewhere else and you connect to them over the common carrier (read: phone company) network. We did this 40 years ago. It was called mainframe computing.

This is an ideal scenario for smaller companies that don’t need much computing infrastructure and who cannot afford a systems maintenance staff. For the SOHO (small office, home office) market, this is a great thing. Over the years I've watched small businesses try to run their own infrastructure and it's not pretty.

But Cloud Computing isn't being marketed as a small business solution. The buzz is that this is the answer for everyone and everything.

However, for companies large enough to have or need an IT staff, there is little value here. What is the CIO going to do? Go to the CEO and say, “I quit. I just outsourced everything to the cloud. My job here is done.”

Let's look at a long-standing example. I’ve always been impressed by the success of Salesforce.com com who, by the way – when they talked to us a few years ago, did not encrypt the data you gave them and put it on a system you share with others. (Anyone read those stories about lost backup tapes...) I can’t see giving the life blood of my company to someone else to manage, nor having my staff rely on the vagaries of Verizon (or, in New Hampshire’s case, the literally bankrupt Fairpoint) as to whether they can do work today. Sorry.

This doesn't mean that we don't use web-based applications. We certainly do. But not for anything critical to our business.

If it sounds too good to be true… it probably is

Your DisplayName here! — Mon, 27 Jul 2009 19:43:06 GMT

One of the hottest topics in the storage industry today is deduplication. Deduplication is this year’s fancy way to do data compression, something that has been available for decades. If you were a fan of compression in years past, then you will probably be a fan of deduplication. But it’s not all upside.

Whereas most of the older compression formats have become standards (Winzip, for example) and thereby readable and writeable by most programs, all of the deduplication formats are proprietary. This means that the only thing that can read the deduped data is the hardware or software that wrote it in the first place. This, of course, is what all the platform manufacturers want – to lock you into their platform.

Dedupe is smarter than ‘ordinary’ compression, which does allow it to compress more, on average. These results, however, come at a price – processing time. Many of the dedupe solutions that perform brilliantly in demo fail when the volume being compressed grows beyond a few terabytes. Their algorithms become unwieldy and system performance deteriorates.

The bottom line for dedupe is the same as it is for most technologies… Under the right circumstances, it offers a lot of value. But dedupe is not the silver bullet. It is not a solution to every problem, and it may or may not be an answer to the challenges confronting you today.

Structural vs. Informative Technology

Your DisplayName here! — Sat, 16 May 2009 16:39:34 GMT

An interesting thing happened this year – an aspect of our business reversed… For years, the bulk of our discussions with customers has been about NTP Software QFS®. QFS provides a structural solution for managing file data. QFS users can set policies that govern the use of their resources. These policies can be both hard (an absolute rule) and soft (warnings, but not prohibitions). But in the majority of situations, QFS users set hard policies. Part of the value proposition for QFS is that it will automatically control what is allowed to go on in your environment.

However, these days it seems that many people have been taken by surprise by the need to manage their storage. Now, all of a sudden, they view storage management as something that has to be addressed immediately – not tomorrow, but today – but they don’t know where to start.

As a result, the character of our conversations has changed. What people want to talk about is no longer about implementing structural solutions, rather we’re talking more and more about how to gather the information necessary to decide what to manage and how to manage it.

The good news is that the applications that can show you where you stand and what your issues may be are non-invasive and easy to implement. You can gather data today, make decisions tomorrow, and start taking action the day after. If you find yourself in a storage management crisis, it only needs to last a matter of days. A non-structural, non-invasive solution is easy to implement and it will provide answers and direction in just a few days.

When it comes to storage management, there is no need to suffer. You can start making progress as soon as the day after tomorrow…

The Fastest Way to Save Money on Storage

Your DisplayName here! — Wed, 06 May 2009 17:19:34 GMT

The speed and depth of the economy’s decline has taken a lot of people by surprise, me included. Today it’s clear that the Storage Industry is no longer immune to reductions in IT spending. While we all still have to keep our businesses running, we also need to cut expenses and cut them now.

Storage, however, remains one of the most expensive aspects of most people’s networks and the end-users’ requirements continue to increase. We need our users to be doing new business every day. This means we’re telling them to create and save new business objects every day. So how do we control this expense? And how do we do it now!

While Harry Potter can just waive his magic wand, the rest of us cannot. Nevertheless, here are some easy to implement techniques that will do the job – and get it done now!

Block all non-business content. There are numerous ways to create file blocking policies depending on your storage infrastructure. The first and easiest step in reducing costs is to stop treating storage as though it were infinite and free.

Set reasonable limits on personal space. There are very few jobs that require the creation of mountains of content. Normally when someone has a large home directory it’s because they never take the time to clean up, or they’re filling it with non-business content. Again, depending on your storage platform, there are a variety straight-forward ways to set limits.

Analyze what’s out there. Unfortunately, there are no easy ways to do this. You will ultimately need find the right application and probably buy a license for it. But a terabyte of storage at today’s prices costs $47,000 a year to own and maintain. Compared to this price, the cost of analysis software is quite modest. For the customers that we work with, their one-year ROI is normally 3 to 5 times the license cost. (Providing, of course, they act on the information the application gives them.)

Everything here can be done on any storage platform. And all of it should have a positive one-year return whether your investment is in staff costs or applications. There is no reason why you cannot make a meaningful reduction in your storage costs this year. None. To quote Nike, “Just do it!”

A Job Offer You Should Refuse

Your DisplayName here! — Tue, 07 Apr 2009 15:07:28 GMT

Over the years a fair number of people have asked my opinion about the job offers they have received. I am happy to say that most of these offers were quite good. In fact, a couple of times I was jealous…

But if I were offered the job of most of the storage managers and CSOs (Chief Storage Officers) I meet, I would refuse. Not because being an infrastructure manager is a thankless job – which it is. Most jobs are characterized by more criticism than praise. The issue for me is that I won’t take a job where I am held responsible for delivering non-stop, high-quality service while, at the same time, I have no control over the consumption of the resources, and little or no insight about how they are being used.

What am I talking about?

Only a few of the storage mangers I meet have any policies describing the appropriate consumption of corporate storage resources. In fact, their users are free to do pretty much anything they please. The majority of these storage managers also lack a system management application that they can use to control the consumption of their storage, nor do they have anything that tells them how the existing resources are being consumed.

Even worse, a surprising number of these people are reluctant to set firm policies governing resource consumption. They live in fear of their end-users.

As an entrepreneur, I’m used to taking risks. But in a successful company, these are controlled risks. I never write blank checks. Why would I want to be responsible in a situation where any one of my users could create a service outage…? And where it’s even somewhat likely that a naïve user will do so by accident one day.

Doing a great job – in storage management or anything else – requires having the right tools for the job. A carpenter without a hammer and a saw is very limited in what he can do. Don’t put yourself in a situation where you don’t have the tools to succeed. Infrastructure management is hard enough even when you have what you need.

This year's themes

Thu, 12 Apr 2007 15:40:04 GMT

Industries and technologies move in synchronous waves. We all experience the same macro-economic cycles, and each of us can only buy what is currently offered for sale. This means that at any given time most companies are in the same place relative to issues in technology. As a result, various periods in time have 'themes' that focus everyone's attention.

This year would seem to have two themes: data security and controlling spending - tough themes to have at the same time since security costs money and generally has no direct benefits to the end users. Even worse, a recent Gartner Group study reported that most IT shops are already under funded this year in the area of storage. This year we have to do a whole lot more with less. The only way to do this is to move quickly and use a two-phase solution. You have to generate an immediate savings that can be used to establish an infrastructure that gives you an on-going reduction in cost.

There are several ways you can take control of your environment and quickly achieve a one-time savings that can be applied to establish the infrastructure that allows for perpetual cost reductions. For example, automatic policy-based removal or migration of data is easy to implement and will give you an immediate break in spending. Follow this with the creation of a self-service environment and permanently reduce your on-going operating costs.

The challenge is to be able to move fast enough in building your infrastructure that you don't get caught by being under-funded. Those who can get it done in time will be the heroes of the next two years.

Beware of Microsoft Vista

Tue, 03 Apr 2007 20:01:02 GMT

Often the lesson is: "If it ain't broke, don't fix it."

Many of us were not looking for a new Microsoft Operating system – I certainly wasn’t. But Vista is here. However, Vista is incompatible with all of the third-party NASes that we have tested against, from NetGear home systems to NetApp Filers.

I can’t imagine why Microsoft would do such a thing, but a lot of people who were open to Vista have since rolled back rather than change all their NASes. As of this writing, (April 2007), I have no information from Microsoft or any of the NAS vendors concerning a solution to this problem or a timeframe for releasing one.

Amazing!

Compliance? What compliance?

Wed, 01 Nov 2006 15:52:50 GMT

Last year Compliance was everyone's hot topic. It's a year later, the noise has died down, does this mean the problem is solved? Not hardly.

We spend most of our days talking to large companies about their storage issues and strategies. Over the course of time, the subject of Compliance generally comes up. What's been done over the last year? Not much. Want to know why? There are at least two reasons for what appears to be a lack of interest in addressing compliance issues - other than in the Banking community, of course.

The first and most important issue is the lack of organization in most people's data. The mere task of figuring out where all the stuff that's supposed to subject to compliance actually is becomes a daunting task in and of itself.

The second issue is what you might expect - cost. Once you find the data you need, you have to rearchitect your systems to bring your handling of the data into compliance. And this expense gives no value to the user community. It is pure cost. If you were in charge, where would you spend your time and money? Not here.

As time goes on and systems get rebuilt, companies will come closer and closer to being in compliance. But today the average company isn't even close.

Santayana's Rule

Wed, 25 Oct 2006 13:29:47 GMT

What was it he said? Ah, yes... “Those who refuse to learn from the lessons of history are doomed to repeat them.” Let's look at some history. IBM created a PC with a proprietary architecture. Others created an IBM-compatible PC that was an open platform. Which one is around today? Apple created a brilliant PC device with a closed architecture and a proprietary operating system. Microsoft created an open OS platform and actively recruited developers. Which one owns the market today?

It has been clear for years (going back to the heyday of IBM) that an open platform and the courting of third-party developers wins the market. In the end, the third parties taken together have more money, get to market quicker and have more ideas than any one company could possibly have, regardless of how rich they happen to be.

Network Appliance and EMC are battling tooth and nail to dominate the storage market. Why then do they both have closed platforms and shun third-party developers? (The one exception being the Centera folks, who seem to have the right perspective.) It certainly seems to me that neither EMC nor NetApp live in a world where Santayana's Rule fails to apply. My bet is that the first of these guys who figures out that if their hardware and infrastructure underly every third-party storage application, and they actually support ISVs, they will rule the Industry. What are they waiting for... someone else to create an open platform and take the market away from them? NIH (not invented here) never wins. It has been tested for years with the same result every time.

Paradigm Shift #1

Fri, 13 Oct 2006 22:18:50 GMT

As the demand for storage grows and technology progresses, the economics of owning storage change. Not all that long ago we wanted to write as much as we could to tape, because tape was cheap and on-line storage was expensive. Now many people think that on-line storage is infinite (almost) and free (not at all, but we'll talk more about this later). In today's economics, storing your data on tape is one of the most expensive things you can do. As a result, many of the people I talk to have abandoned tape altogether for on-line archives. Not only are the on-line archives cheaper, but they are more flexible and much easier to search. But this doesn't mean you can manage them like tape...

With tape the deal was write it and forget it. Tapes in a mountain somewhere generally receive no maintenance (not that this is the right thing to do). On-line storage, on the other hand, needs maintenance. As you write more, you need to provision more. And if you don't organize it well, as you write more, specific things become harder and harder to find.

The bottom line is that this is a real paradigm shift. It is not as simple as substituting a disc for a tape. You have to manage differently as well. It's time to change your thinking as well as your media...

DIstinguishing between health and risk

Tue, 10 Oct 2006 13:52:14 GMT

Storage is just bits on a disk, right? What's complicated about this? A lot. Storage is a surprisingly complex thing to manage.

I tell you that a volume is 95% full. What does this mean? I tell you that another volume is 70% full. What does this mean? Maybe not what you think...

A volume that is 95% full is unhealthy. There is no getting around this. The algorithms that allocate space on volumes start breaking down after 85% of the space is taken. Fragmentation begins to increase dramatically, and there is not enough free space to reorg the data in place. In other words, the volume is unhealthy. The reaction of most people to this situation is to say that this volume demands immediate attention. Seeing only 5% free space, many a wise storage manager would declare an emergency.

But suppose I also tell you that the volume's size has been stable over time. It neither grows nor shrinks by more than 1%. Now there is no immediate emergency. The volume in question is unhealthy, but not very risky. Should you apply your efforts here, or do something else?

Our other volume is 70% full. At 70%, we know this volume is healthy. But now suppose I tell you that the space used is growing by 10% to 15% a month... This volume is certain to run to be completely out of space in less than 3 months. Sounds pretty risky to me. I might put my attention here and let the other volume slide for a while.

The point of the story is that things - storage management in particular - are rarely one-dimensional. If your model for storage management is over simplified, you might make serious mistakes, or you might be constantly chasing your tail never knowing what is truly important. As you plan to manage your storage cost-effectively, you need more information than just performance and free space. If all you know is performance and free space, your only solution to any issue is add more hardware. Hardware vendors love this solution, but your senior management doesn't. Nor should you.

What happens on the last day?

Tue, 03 Oct 2006 14:21:03 GMT

Many companies have dropped tape in favor of on-line archiving. This is certainly a good economic strategy and a requirement in some cases. But what's the end-game? You can storage an infinity of data on tapes and throw them in a mountain somewhere (data waste to go along with nuclear waste) at a low cost. (It's writing the tape that is expensive.) But the bits that go to the on-line archive simply require more and more storage... Who is doing the math? This is an exponential curve. Remember, for most organizations, the need for storage is doubling every 12 to 18 months (Gartner Group). This year you added 500 terabytes... Next year you don't add 500 terabytes, you add a petabyte.

How many doublings does it take before the cost of storage exceeds the company's annual revenue? I've yet to have anyone give me their plan for 5 years out. What's yours?

How much does junk cost?

Sun, 01 Oct 2006 20:51:29 GMT

We've heard it over and over again. Gartner Group tells us that 25% to 40% of what's on our storage is junk - expired, unneeded, inappropriate. What are we doing about it? Mostly nothing. However, as storage expands and companies move to tiered storage, the cost just grows and grows. First we pay for it on primary storage. Then we pay to migrate it to secondary storage - and the space to keep it there. Finally we pay to move the junk to archive. So we don't just pay for it once, we pay for it 4 or 5 times over.

Given this, the value of getting rid of even a fraction of the junk isn't a few terabytes, it's a few terabytes times 4 or times 5. This is real money.

How do you back up 750 terabytes of data?

Tue, 14 Feb 2006 21:06:58 GMT

Recently in London I was talking to the CIO of one of the larger Brokerage firms. He noted that they currently have 750 terabytes of storage and that he expects by this time next year to have twice as much 1.5 petabytes.

How do you backup 1 petabyte of storage? The answer is: you don’t. At least not in any conventional way… there is no technology on the market that can write a petabyte of storage to off-line media in any reasonable amount of time. Size changes everything.

Now if you’re a Global 2000 multi-national like this brokerage firm, you may be able to skip backups altogether. (Backups, not archiving) If you have a world-wide distributed database, say in London , New York , and LA, and all of London , New York , and LA have ceased to exist, you probably don’t care very much about those very safe tapes in Iron Mountain … There would be no place to restore them to.

This is a turning point in our industry. Conventional backups must now give way to on-line backups, tiered storage and archiving. Don’t go buying stock in any of the tape companies…

Our naked emperor - data security

Thu, 12 Jan 2006 17:12:44 GMT

We all probably remember the parable about the emperor who had no clothes, riding around naked while his whole kingdom pretended that he was wearing the most beautiful set of robes... Today we have about as much data security as he had clothes. Early last year Mastercard admitted that a backup tape, written in clear text, with about 2 million customer records was lost. Last month (December '05), Marriott revealed that a tape with a couple hundred thousand customer records is missing.

Who among us thinks that these are isolated, once in a life time events? Not me! My organization has lost tapes before. Humans make mistakes. You have to assume anything that gets handled by humans can be lost or damaged. And the more humans that handle it, the more likely it is to be lost or damaged.

It's time for the industry to get it. There is little possibility of real data security until all data is encrypted. Until then, we're all just pretending...

One topic from the CES

Wed, 11 Jan 2006 23:15:31 GMT

The Consumer Electronics Show in Vegas always provides plenty to think about. One of the more interesting discussions concerned putting hard drives in cell phones.

I remember the day - years ago - when I figured out that my laptop would eventually have a gig of memory. I thought that was pretty wild. Never did I consider that my camera would have a gig of memory. (It does - about 400 pictures worth.) Also never did I consider that my phone might have a hard drive... This changes everything. What would it mean to have - and I mean all - of your data in your cell phone? What will it mean to Corporate America that everyone will be coming to work with the ability to download and transmit pretty much any and all information they can access?

And how are you going to backup your cell phone? The data there won't be mirrored on network servers. What happens if it goes away?

To date the phone / PDA has been a pretty klunky device and not all that useful. This is about to change, and when it does, a lot of other things will change.

Built right, or built wrong?

Wed, 21 Dec 2005 19:21:56 GMT

As technology strategists, it is our job to select the right technologies that will accelerate our companies’ success in their chosen businesses. Often a single poor choice will cause our whole project to fail. Sometimes even whole companies fail over the technologies they chose to invest in. (Where is IBM in the PC and networking business now?)

Why, then, do so many companies use a second rate process over and over again to make their choices? Why does senior management encourage it to happen? It seems to me that the right way to make choices isn’t all that hard to find. And, by the way, this isn’t limited to picking technology; it applies to most decision making.

Let’s look at what generally happens. Most organizations build a matrix of features, recruit a few vendors, fill out the matrix, give a high weight to cost, and make their decision. Sound familiar? How well does this correlate to what most companies state as their mission? How many companies do you know whose mission statement is: “We want to be as good as we can be provided doesn’t cost too much”? Or, “Our commitment is to excellence, unless it’s expensive, in which case we might not do it at all”? Is this the way we really run our businesses? In some cases, I believe that it is.

Where I work, we run the business differently. We use a sequential process designed to produce the best possible solutions (and the best possible business) every time. Here are the steps:

Step 1: Does this problem really need to be solved? There is no lack of things for us to do. If the need isn’t compelling, then we shouldn’t work on it at all. But if we resolve to work on an issue, then it becomes policy. The problem will be addressed.

Step 2: Is this the most important thing for us to do next? Look at the government. The reason things are such a mess is that there is no rational order to the way things get addressed. Time and resources – not money – are our most precious possessions. If you allow them to be spent in the wrong order, you can get lots done and still fail.

Step 3: What is the best possible solution? Here is where everyone needs to bear down. Why is this the best possible solution? What alternatives did you consider? What values are you focusing on when you choose this one? Most people do a weak job here. Management doesn’t want to press too hard (and alienate their staff). Staff members are often reluctant to take clear cut positions and be right or be wrong. In our organization we simply force people to do this. It’s a necessary and required ingredient for success.

Step 4: Can we afford it? Which one of us comes to work intending to be mediocre? Who among us plans to be second rate? If you really have identified the best possible solution, why would you want to do anything else?

Everyone knows how to spend money. If you start with a budget you get a plan to spend it, not a solution, and guaranteed, not the best solution. If the way you handle cost is other than “Can we afford the best possible solution”, then, by design, you have committed to build a mediocre institution. Is this what you really want to do? Would the CEO give you a raise if you told him this was your plan?

World-class organizations are the best at what they do because they engineer mediocrity out of their institution. What does your company want to be?

Where's the money?

Wed, 14 Dec 2005 14:54:45 GMT

Remarkably, as we work with clients we discover that they rarely analyze the component costs of their operations. So let me make this easy: the bulk of the money is in whatever you do for data protection. But even if you know this, have you looked at the details of what you’re protecting and how it flows through you systems? What components make up the bits you write? How often do you write them and why?

Last year we worked with a client who had done this analysis. They discovered that 40% of what they were writing each day was user PST files in home directories, independent of server backups. (This is a financial institution whose corporate policy requires the preservation of email. Each user needs and wants their own record.)

Outlook, as you may know, has the annoying feature of updating your PST file each time you open it whether you have changed its contents or not. This means that if you do daily incremental backups every PST file for every user is written every day. And, if you’re keeping most or all of your emails, and you have lots of users, this is lots of data. Can this be made less burdensome and less expensive?

We worked with the client to set a system-enforced limit on the size of an individual PST file. This means that each user now has a series of smaller files. Outlook only attaches and updates the current PST file; the others are left as they are. Enterprise-wide, this eliminated 100 tapes from a year’s worth of incrementals. 3M is commonly cited as stating that the cost of maintaining writing and maintaining one tape in a typical backup scenario is about $3,500 a year. One hundred tapes times $3,500 is a lot of money – a third of a million dollars a year. For the sake of argument, let’s assume that your operations are less expensive – only $1,500 per tape per year. This is still $150,000 a year.

We would all agree that every user’s archival mail is something we have to protect. But, maybe it’s worth drilling deeper into how you do it…

The new economics

Thu, 01 Dec 2005 10:16:11 GMT

We have all watched the price of on-line storage decrease year after year. While it is still neither infinte nor free (and it's certainly not free to manage), it is pretty inexpensive these days. So inexpensive, in fact, that tape is now dramatically more expensive. This cost inversion (tape used to be much cheaper than on-line storage) should change the way we manage our storage.

If you are a large corporation with sites all around the world, you need to replicate your data to these sites anyway. If London, New York and LA have all ceased to exist, you probably won't care much about the tapes at Iron Mountain... Trust me, your concerns will be elsewhere. As a result, there is really no need to have any of these tapes.It's a costly security blanket that really accomplishes nothing.

A smaller company needs tapes for disaster recovery, (although on-line vendor-based alternatives exist) but daily backups should be on-line. Writing tapes everyday is error-prone and costly. You just shouldn't do it anymore.