General discussion


The Myth of Scalable Technology Solutions

By dcolbert Contributor ·
For a long time, I and most people on the planet have seemed to realize that consumer/home PC scalability is largely a myth. You certainly CAN upgrade that system just by purchasing a new CPU, removing the old CPU from the ZIF slot designed to make it ever so simple to do, and putting in a newer, faster CPU... but it happens so very rarely that it is hardly worth mentioning. Quite likely, if you buy a lower end system with the intent of upgrading it later, you'll find that by the time you are ready to move up, the entire platform of technology has moved on, and the best you'll be able to manage is a compromised solution. There is an odd paradox here, as the emerging bleeding edge often has quirks, incompatibilities, and non-standard features that are likely to be worked out a few short months down the road in mid-range equipment that costs far less than the premium price tag paid for bleeding-edge tech.

In a nutshell, buy the PC you can afford that will meet your requirements, use the hell out of it, and later on, once it starts to show age, buy a whole new one. By the time you upgrade your DIY system and find that you now need new ram, a new SATA drive, a new video card in a different slot standard, etc, you're almost always money ahead just to go with a new machine.

I'm an avid DIY builder, and I do upgrades all the time. Some memory here, a bigger hard drive there, a new video card. I've also built a wealth of knowledge about the shortcuts and contacts that save money and headache when doing an upgrade. But mostly, I do it as a hobby. There is a good analogy to automobiles here. For the average person, building a kit car or a hot rod or a custom car just doesn't make sense. You buy a new or used stock car, you drive it as long as it suits you, and then you get rid of it and get a new one. Automobile hobbyists often spend as much if not more than they would have to just go buy a stock car with similar characteristics new. The hobby is what drives them (and perhaps the ability to "pay as you go"). I really think the DIY PC market operates on similar principles.

But, recently, I've experienced something new but related to this, in the enterprise equipment arena. Shockingly, I've never really heard any mention of this in forums like these or in industry publications, nor have I heard my peers mention it. It seems to be the Pink IT elephant in the room, almost.

Are scalable enterprise technologies a myth? I've had three examples recently, all through Dell.

We had a Dell 132T tape library with a DLT2 single drive when I arrived. I wanted to add a second drive. The literature and sales brochures all pushed the scalability of the solution, in being able to add a second drive later when your needs required such. Unfortunately, a couple short years, maybe less, than the original purchase of the device, and Dell no longer offers the second drive.

Next, we went to upgrade our Dell EMC CX300 SAN with a new fully populated enclosure. While this was available, it was conditional. The new technology is at 4gb/sec, but of course, our older CX300 storage processor and enclosures are limited to 2gb. It also required an outrageously expensive and intrusive Flair OS upgrade. The promise of ease of scalability simply wasn't delivered on. It was far more expensive and far more headache than SAN literuatre would lead you to believe.

Most recently, we purchased Power Edge 1955 Blade Servers and enclosures this last spring. This is the only purchase of the three that has occured since I've been here. Again, like a SAN, the primary sale point of blades is the ease of scalability. Yet, the blades themselves have already been discontinued. Let me elaborate, prior to these blades, we had another, older blade enclosure populated with blades. We had to buy THIS blade solution because that earlier one had been discontinued.

It seems to me that this is a similar effect in action to the DIY "upgrade" cycle. Trying to buy "just what you need now in a scalable, easily upgradable" product is a fool's errand. The scalable, easily upgradable portion is going to be hopelessly out of date, if even available, by the time you are read to move forward. The concept of modular, scalable upgradability in enterprise components is a myth. If you need a SAN now, fine. If you want the benefits of blades now, fine. But don't let that sales-pitch that, "And if next year, you need to grow, it will be a piece of cake to add another enclsoure and add more components to meet your needs". If you need more a year or two down the road, you're VERY very likely to have to throw that old baby out with the bathwater, to waterfall the old stuff into some less critical area of your enterprise and replace it with the latest, newest and most shiney incarnation of whatever it is that you need to "scale".

What are your experiences? Have you purchased enterprise equipment on the merits of long term scalability and seen that promised delivered on, or do you share the same experience as me?

This conversation is currently closed to new comments.

Thread display: Collapse - | Expand +

All Comments

Collapse -

'Scalability' vs. 'upgradability'

by CharlieSpencer In reply to The Myth of Scalable Tech ...

It sounds to me like you're discussing being able to upgrade hardware. I've only heard 'scalability' used to refer to software continuing to be useful as the data load grows.

Collapse -


by dcolbert Contributor In reply to 'Scalability' vs. 'upgrad ...

I hate to get caught up in a symantic debate - but I suppose it is important that we are clear.

Nobody in the industry pushes San solutions as "upgradable" they push them as "scalable". So, I'm using the industry term as I understand it in that sense.

Scalable - the ability to grow to meet the growing needs or requirements of your IT organization. "Up and Out growth" - or other industry buzzwords. Generally followed by the pitch, "You only need to purchase what you need now, because this solution is scalable, you can grow it as your organizational needs grow".

Nobody "upgrades" a blade server solution, they "scale" it. But effectively - it is interchangable with the idea of an "upgrade".

"I bought the PC I could afford that met my needs today, but tomorrow I plan on scaling(upgrading) it by adding a PCI-X video card and a couple of gb of ram".

And in both cases, all the caveats I mention that prevent DIY hardware upgradability from being entirely practical affect the enterprise as well - called scalability or upgradability, it seems to be two spins on the same concept.


Collapse -

I don't want to debate symatics either.

by CharlieSpencer In reply to Hmnnn

I only saying I've never heard 'scalability' applied to hardware. That doesn't mean a sales puke hasn't used it with you.

I think it's like 'hypoallergenic' as a technically undefined term used by marketing types to sound knowledgeable. I suggest if someone uses the term, press them hard for concrete examples of how the technology scales up, and what are the upper limits of it's range. Keep pushing until you either get hard limits or until you find a vendor that doesn't use vague terms.

Collapse -


by dcolbert Contributor In reply to I don't want to debate sy ...

(Now, the question is, do I play it off as if I didn't spell it "symantecs" or do I confront the mistake directly... hmmm... bring on the active denial!)

So, anyhow... well, I think the problem is that the limits of scalability are well defined for the product at that time. The CX300 is a good example of the limitations of the scalability being well represented.

The problem is, it is the predictive, forcasted limitations of scalability that are really impossible to accurately predict that are at play here. In technology, the odds are that you'll lose on the hardware scalability game, unless you plan to scale VERY rapidly (in which case, you're probably better to jump in with both feet at the start rather than start smaller and scale, anyhow).

If you buy a SAN solution now, starting small, with the plans to grow and scale later... what are the odds that SAN technology is going to stay pretty much static without major improvements and changes in the next 2-4 years? Or Blade Technology, or any other technology? By the time you decide to scale, at best, you may be sinking money into technically "obsolete" and dated equipment when newer, more reliable, feature enhanced and less expensive equipment is available, at worst, you may find your technology effectively orphaned by your vendor.

I agree, it is mostly marketing smoke and mirror hyperbole designed to give a "Value Add" to the product. But it seems like it never gets addressed in the enterprise technology sector, whereas it is pretty much recognized in the consumer sector. I find that part odd.

Collapse -

Do you burn your money in a fire?

by ---TK--- In reply to The Myth of Scalable Tech ...

"buy the PC you can afford that will meet your requirements, use the **** out of it, and later on, once it starts to show age, buy a whole new one"

How about stay 3 years behind technology, and build for the future.... "once it starts to show age" beef up the HDD's, reformat, and turn it into a file server... and start rotating your equipment around...

Collapse -

I'd rather turn a three-year-old server into a PC

by CharlieSpencer In reply to Do you burn your money in ...

than turn a three-year-old PC into a server.

Collapse -

on a corp level....

by ---TK--- In reply to I'd rather turn a three-y ...

is a different story, but on a personal level its a little different... for me, I took my Socket A (Yes it still runs) :) slapped 2 TB's and turned it into my FTP / print / torrent PC . For me, it doesn't need to be lighting fast, I just need it to be stable.

About every 2 months it needs to be rebooted... works great for me...

Collapse -

This approach

by dcolbert Contributor In reply to on a corp level....

I've got a stack of "desktop grade" PCs running various "enterprise applications" in my datacenter - and it was that way when I came here.

I have my doubts that this kind of frugal "waterfalling" of equipment is practical or advisable. I see SOHO consultants doing this kind of thing for small businesses - and I see the same kind of guy who becomes a SOHO consultant do this at home with some Frankenmachine Linux beast that he uses for a FTP/Print/Torrent geek server - but at some point, from a business perspective, having a stack of P3-1ghz desktops leaning in your datacenter says something about your focus on delivering a true enterprise class environment.

From my perspective, I would like to EOL that "tower of PCs" and move the individual roles of each of those PCs onto virtualized machines running on a single server classed machine or blade. I think the benefits of supportability, of having enterprise class support and engineering, outweigh whatever short-term "benefits" the original architect of my data center gained by using desktop grade PCs in his enterprise datacenter.

Now, with that said, there is nothing wrong with waterfalling enterprise grade equipment through your data center. As machines fall of the technical edge there may be application for them in non-critical roles, in test and dev or otherwise contributing to your enterprise.

But we're talking about something completely different here. We're talking about the marketing spin behind things like "blade" servers.

One of the pitch of blades is that you can buy an "unpopulated" blade enclosure, buy as many blades now as you need, and add blades LATER as your business expands.

The practical reality is almost inevitably that by the time your business expands and you are ready to add those blades, your vendor is going to tell you, "We no longer make that enclosure or those blades, you'll need to buy a NEW enclosure and NEW blades".

Collapse -


by ---TK--- In reply to This approach

Yeah, like I said a Corp. environment is a little different. It really depends on what the stack of old dells are doing. The only way I see them fit for is a Linux computing cluster... Anything short of that they wouldn't be in the data center. All they do is bang out jobs...

"Linux beast that he uses for a FTP/Print/Torrent geek server" LOL... if you are attempting to take a shot at me, I am failing to pick up on it. You are assumptions are a little off. First of its not a Linux box, its w2k3, second... look at the minimum requirements for w2k3, What I am doing it doesn't exactly require much muscle... third my linux box is my proxy/firewall...

"We're talking about the marketing spin behind things like "blade" servers." Yes I am fully aware that the main point was about blade servers. My main underlying point was to stay a little behind "the latest and greatest(hardware)", buy in bulk which will save you money in the long hull, and redistribute old technology.

Collapse -


by dcolbert Contributor In reply to haha....

Not really taking a shot at you - I'm taking a shot at US. :) I mean, we're nerds, we can make the barest of machines do wonderful roles and tasks and that is often why we're in these roles. Someone has seen what we can do with this technology and has decided that we deserve a job, at some point.

But *some* of us have trouble drawing a distinction between the clustered, distributed data center running on old AMD K2/400s running in our basement and what is suitable for a corporate data center. Most of us, I think, are prone to go, "I can make this happen for a 10th of the investment or less if I just..."

But when you go that route, you're in dangerous waters. At home, for a small CPA office with 4 employees maybe, you can get by with these solutions. But when you've got a business with significant revenue handled by your corporate IT systems, you don't want a 7 year old P3 1Ghz system to disrupt any part of your revenue flow.

If I seem insistent on bring this back to focus, it is because I'm really curious. It seems like the IT industry doesn't believe that this issue of scalability being a non-viable option in general applies to it.

Related Discussions

Related Forums