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.

Tuesday, October 10, 2006 9:52:14 AM (Eastern Daylight Time, UTC-04:00)      Comments [0]  Storage Management

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?

Tuesday, October 03, 2006 10:21:03 AM (Eastern Daylight Time, UTC-04:00)      Comments [0]  Observations | Storage Management

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.

Sunday, October 01, 2006 4:51:29 PM (Eastern Daylight Time, UTC-04:00)      Comments [0]  Observations | Storage Management

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…

Tuesday, February 14, 2006 4:06:58 PM (Eastern Standard Time, UTC-05:00)      Comments [0]  Observations | Storage Management

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...

Thursday, January 12, 2006 12:12:44 PM (Eastern Standard Time, UTC-05:00)      Comments [0]  Management | Storage Management

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.

Wednesday, January 11, 2006 6:15:31 PM (Eastern Standard Time, UTC-05:00)      Comments [0]  Observations | Storage Management

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?

Wednesday, December 21, 2005 2:21:56 PM (Eastern Standard Time, UTC-05:00)      Comments [1]  Management | Storage Management

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…

Wednesday, December 14, 2005 9:54:45 AM (Eastern Standard Time, UTC-05:00)      Comments [0]  Observations | Storage Management

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.

Thursday, December 01, 2005 5:16:11 AM (Eastern Standard Time, UTC-05:00)      Comments [0]  Storage Management

Listening to all the hardware vendors, you would think that virtualization is the greatest thing since sliced bread... It is true that it solves a number of problems, but it also comes at a price. For example, many organizations maintain historical backups. Suppose I use the facilites of my virtual storage to do some reoganizing. How do I then trace the files that I have - in their new locations - back to their ealier backups? You can't. At least you can't without NTP Software's technology. Does it matter? It matters to a lot of our customers. In fact, we work with clients who refuse to reorganize their storage for just this reason. The bottom line? Everything comes at a price...

Thursday, December 01, 2005 5:03:55 AM (Eastern Standard Time, UTC-05:00)      Comments [0]  Storage Management