Slashdot Mirror
How Do You Keep Up with Enterprise-level Tech?
E1ven asks: "I'm curious how the Slashdot gang chooses to keep up with the performance of high-level equipment for servers, routers, loadbalancing, and the like? For PC-type specs it's easy, every guy and his dog has a review website, and magazines stuff themselves in every window. However, the higher-end equipment is far more difficult to find trustworthy analysis of.
I'm curious how other people have solved this problem, and what resources they use to keep on top of the game?"
I buy what the vendors tell me to buy. After all, they know what's best, right?
Interested in open source engine management for your Subaru?
Get a tech job at a major enterprise and let them educate you on the subject. It worked for me...
oh and some friends, some of the higher end systems are on the higher weight end as well.
Depending on the environment you are in, you aren't making significant purchases that often. When you are, it's usually because you've outgrown what you have and can't simply N+1 it anymore. Either that, or it's tied to a brand new product offering. So you do the dance:
1) What are my competitors using?
2) Do any of my current vendors have a solution, and it is worth it?
3) Who is number 1 at the technology I'm interested in, and why?
4) Am I going to need contractors for initial implementation, or is the talent for this technology in house?
5) What's training going to be like?
Then you do a whole lot of research and select vendors(s). You let them come out and do a presentation if that's appropriate. Nine times out of ten, you'll end up going with the proven solution that a lot of people are already using. It's easy to make a business case for a known quantity.
Unfortunately, that's not how it usually works out. Other things color the decision like:
1) This friend of mine still works at this company and I'd like to throw them a bone.
2) For political reasons, we like company A.
3) The upper management prefer product C because of the pretty colors, and because so-and-so heard it was great at some cocktail party.
4) We are going to use solution D and that's official from upper management. There is no discussion. They read about it in CIO Monthly.
5) I have stock in company E.
You get the picture.
For every annoying gentoo user, are three even more annoying anti-gentoo crybabies. Take Yosh from #Gimp for example.
Keep up with Enterprise-level Tech? Mostly by watching Star Trek of course! Duh! :)
The Register has a news section on enterprise computing. I wouldn't say that's all you'd ever need to read, but its a start.
http://www.theregister.com/enterprise/
Consult the BOFH series of articles on how one keeps up. It's on theregister.co.uk Sure it's dated in some places but a good read. Other then that vendors are willing do bend over completely to get a sale. I recently got a Demo of Weblogic from BEA and on for Webtrends. Just for asking. If your a serious (or even not) buyer, they will give your almost everything you need.
Sorry about the writing. Robot fingers, you know? Cliff Steele in DOOM PATROL #23
Not always available or appropriate, but you do your research and take the plunge--but before you put all 20,000 users on the chosen solution, you try it out with 100 and then scale it up as it is proved.
The problem with this is that some products are only available to WAN type solutions, so it's either 20,000 users or none. For those, it's naturally more difficult; who do you ask about how the thing is going to work? How many other institutions use the exact product you're investigating, in an environment that's analogous to yours, who also isn't a competitor and therefore not willing to reveal their competitive advantages?
For those, I'd say: do your research as others above have suggested, but then it comes down to nailing the vendor on deliverables--what are the consequences of the product not performing as promised? Rebate, return, free upgrade to the bug-squished version? And support contracts--how much support time does the solution come with to make it work as advertised?
If anyone who wants to sell to the Enterprise isn't willing to give you both written guarantees as to performance--and consequences for failing to perform--as well as some support, they're not really ready to be an Enterprise vendor. That's part of what that $1M buys you.
--
$tar -xvf
Ask for whitepapers, documentation, a sales call... presentations, etc etc
If you're planning on buying something that is... otherwise.... hmmm why, it will all change by the time you need to make a decision.
A fool throws a stone into a well and a thousand sages can not remove it.
So, like, transporters?
Two of the large corporations I've worked at paid a bundle to have access to the Gartner Group reports. When we were doing tech evaluations, the people who had access to the reports (you pay for search ability) would then sort of summarize things for us, or in some cases, they'd pay a fee to Gartner so that the rest of us could legally read the report, or illegally just email the rest of us a PDF and tell us not to tell anyone.
But, basically, you have to find someone who 1) has the money to do the research, 2) has the experience to make quality comparisons, and 3) is willing to do it in an unbiased manner.
Sometimes large vendors (like IBM) have things like a "Customer Care Council." These are vendor-supported forums for the customers to exchange information, but there's fees involved, or NDA's that the Corporations have to sign.
I've also come across some companies that pay for research. They get evaluation or minimally licensed copies of the software, then try to run a comparison. The information I've seen come out of these is usually heavily compromised by someone with a bias. For example, one corporation I consulted with earlier this year paid $3 million for research on what to upgrade their POP email system to. The result? I'll give you a hint: the company writing the report also sold hardware that only ran one operating system, so they naturally supplied a report putting that operating system (and what would run on it) as the choice -- and supplied price quotes for the hardware using only their hardware.
So that company paid $3 million to switch to a more expensive email solution. Doh!
and ask for demo models. The company I work at does that all the time. We just purchased a couple new Redline load balancers that way.
Cyberbite Networks - Web Hosting, Dedicated Servers & Colocati
When you're talking about real money, you do what research you can to scope your requirements, then get some eval units and test the hell out of them with your application. Repeat for each vendor under consideration, then decide.
It's called due diligence. Just Do It.
-Isaac
I am not a lawyer, and this is not legal advice. For Entertainment Purposes Only.
welcome our new Opteron-reviewing dog overlords.
Monstar L
This is what I do to get my enterprise running smoothly:
* Level 5 diagnostics every hour
* Level 3 diagnostics on first sign of battle ready
* Level 1 diagnostics once a year
* Inspection of warp coils for tetrion or verteron particles. These can cause poor engine performance.
If your enterprise is run on clusters, then what you keep on is mostly clustering software rather than high-end hardware, the exception being the switch fabric. So keeping up with enterprise-level tech for me means understanding things like Oracle's Cache Fusion / 10g and then knowing enough about tieing it together, which is really a network/SAN thing. For the server hardware itself, off-the-shelf stuff from IBM/Dell/HP all work fine.
http://www.gartner.com/
http://www.metagroup.com/
http://www.idc.com/
http://www.forrester.com/
http://www.idg.com/
http://www.jupiterresearch.com/
http://www.yankeegroup.com/
http://www.aberdeen.com/
http://www.amrresearch.com/
And yes, they all cost money. If you're an enterprise and you want input on how to spend you tens-of-thousands to multi-million-dollar IT budget, you can shell out a few more dollars to get some research.
In companies where the tech team drives the tech, I have seen a group of computers set aside for experimentation. They won't spend too much time on it, but from time to time, they'll try something out to see if it is promising.
If you can find a company where mangement listens to tech, it is usually pretty easy to justify this experimentation lab environment. You still have to show how having it is better than not having it, and how much it will cost over the next three years. But that's no different than any other project in techland.
Companies where the upper management drives tech decisions are generally not the best places to work. I tend to avoid them. I guess I've been lucky because of my particular micro-field (e-commerce perl programmer), while others I know always end up in top-down organizations (java programmers).
The radical sect of Islam would either see you dead or "reverted" to Islam.
The point is that they'll let you take a fairly in-depth look 'under the hood' before you buy. I mean, you pay enough for these boxes/software that if they didn't they'd be shooting themselves in the foot.
As far as knowing which vendors to evaluate, we stay up to date by reading Slashdot
Dealing with Enterprise tech? Easy. When all else fails...
REVERSE THE POLARITY!
Oh, wrong Enterprise...
Y'know all those weekly and biweekly glossy mags your CIO and Manager get? There actually is valid product overview and release information in them. The IT executive industry base is a PR flaks paradise. Read/scan and then scurry to the vendors website for the White Paper/lowdown.
If you have the money, set up a test lab. Vendors are more wary to loan eval hw than they were in the dotcom era but a relationship with a good salesman can always finagle some product.
Have the suits do the talking if you are buzzword intolerant. Warning: suits talking will often lead to mutual masturbation, technical direction by fiat and the resulting stillborn initiative pandemic.
Envy my 5 digit Slashdot User ID!
Long ago, pre-web, there was a great service called "Computer Select". You'd get a CD-ROM every month, with the text of most of the last year or more of the industry rags on it, and a very good (DOS!) search interface. So if you were researching Ethernet switches, you could get *all* the magazine reviews and articles in one place. It was a fantastic purchasing resource -- I saved my company jillions and knew about the second-tier manufacturers that had better products. I found some suggestion that it was being revived, and even found the "replacement"'s web site, but got no response when I tried to fill out their form or when I called for an eval. And no, Google is not a replacement -- even if you wade past the first three pages of SEO-enhanced "Buy Here!" sites, you still have too much crap to wade through and still miss "real" reviews and articles (and some mags' texts aren't accessible).
Where I'm at we have a very simple rule. Many people don't like it, but it's the way things are. Since I rule, it's all good.
1) Switchs and Routers are all Cisco devices. Using Cisco NNM to monitor, configure and control them.
2) All servers, printers, Pc's and laptops are all HP. All controlled by Insight Manager.
3) All UPS are from APC and we use InfrastrXure manager (everything has a SNMP card) to make sure 1 & 2 are powered on. Keep it simple, don't mismatch products.
... buy IBM!
What do you need? Mainframes, mid-range, Unix or Linux servers, laptops, storage systems, enterprise printing systems, middleware, systems management software, consulting services and a partridge in a pear tree.
Full disclosure: Yes, I'm happy to be a blue drone.
It's all you need, really
The perfect sig is a lot like silence, only louder
There's no easy solution (hey, if there was, they wouldn't need to pay you.) Generally you'll do just fine if you follow a few simple guidelines.
1. Make damn sure you fully understand your own needs. A mistake here is going to get you serious butthurt in the long run. On the other hand, if you really understand what you need, the products that match kinda jump out at you. A lot of folks run off half cocked chasing new toys, then find themselves working like dogs to implement a whizz-bang solution that simply doesn't fit the puzzle. That's not the way to go.
2. Identify potential solutions. First, talk to your trusted vendors. If you have good ones, they can give you a quite thorough grounding in the subject. Make use of google, dig up user testimonials, case studies, news articles. Talk to your other IT friends. I get a hell of a lot of recommendations through them. Post an Ask Slashdot. Check research analysts. Build yourself a list, preferably longer than five.
3. Everyone has free evaluations. If they don't, call them on the phone, and they'll send you one. If not, pirate it to evaluate it. It is lunacy to use a product you yourself have not yet fully tested in your environment. Build yourself a lab of a couple computers, and test the hell out of your potential solution. Let the tech-savvy users in your organization (your go-to guys in any given department) test the user end. Continue until you find the product that meets your needs. This is also a good time to break things really spectacularly and then call up the vendor's support lines and find out what their support staff is made of. If you get someone who is reading a script (especially with an Indian accent), hang up and switch to another product immediately.
4. Make sure the product will take care of you in the future. Everyone skips this step. Make damn sure the company isn't going to vanish overnight, make sure they provide quality support, investigate your options for warranty extensions and maintenance contracts. Verify that they plan to continue support for all of your needs in the future. Take a look at their future product roadmap.
5. When you actually implement your solution... document it. Even if it's a badly typed word document that sketchily outlines what you did. It takes very little time, and will make life easier for you (and everyone else) in the long run.
6. Profit!
Hell is being intelligent in a world full of idiots.