Archive for the ‘storage’ Category
It’s on: I fly out to Denver, the Mile-High City, on Tuesday. Three intensive days of TTT (Train The Trainer), a semi-social visit to another company site, and then a few days vacation in and around Denver, Colorado.
I would call it a “holiday” if it was in any other country. I change planes at Heathrow, which is risky, but the alternatives were Frankfurt (totally unfamiliar to me), or various US airports, but I’ve been warned against going through any major US hub.
I was only granted permission to go on this course, on expenses, on the grounds that I will redeliver training on the product family. Later this year, in Bangalore is the current plan; or it was the plan a week ago. Since then something has happened that blows the situation wide open, and my involvement could go either way. I’ll explain as best I can without exposing any sensitive information.
OEM is short for Original Equipment Manufacturer, and applies to any company who makes a product released under another company’s name. Dell, from example, does not actually make anything at all: it is an assembler of PCs filled with OEM components from such companies a Intel (CPUs), Seagate (hard drives), or Broadcom (networking).
The OEM company behind my trip to the USA makes storage software, which my employer re-sold under its own name. It’s fairly high-end stuff, but it can take advantage of our less-expensive hardware in a horizontal scale-out architecture, so it has the potential to become even bigger than it is, by entering the “midrange” market aggressively. We’re due to meet company representatives, and be trained by them on advanced diagnostic techniques and system planning.
In the middle of last week, however, it was announced that my employer (a big company) is acquiring the OEM (a small company). The employees of the OEM are becoming colleagues of mine.
I can see two possible outcomes of this. On the one hand, the barrier between the two companies is being torn down, and I will have more direct access to knowledge and resources. It could mean that support for this product is no longer needed, and I go back to other products.
It could also mean that my role expands, since my employer has just demonstrated its full support for the technology in no uncertain terms, and probably plans to roll it out on a wider scale. Which reminds me of what Victor Kiam said about Remington, all those years ago: “I liked it so much… I bought the company.”
There’s been very little happening on the home front since my return from points East. It’s been almost four weeks, filled with little more besides work, sleep, and watching the final series of The West Wing.
Of The West Wing, all I will say is: those critics who claim the quality declined in later series’ are blowing bubbles; sure, things changed, quite radically, but it reflects the reality of what happens in the US Government (Executive Branch) in an Election year. Each Presidential term is a real roller-coaster ride, from which only some of the characters climb off at the end, having earned the privilege of walking away in the certain knowledge of a job well done. Highly Recommended.
It’s happened before that I’ve gained new insights on a topic by writing about it here; it’s at least as useful to talk about it, to a bunch of people who may have a different take on the subject. That was the case last month, when I gave two weeks of consecutive training courses to my company’s new recruits in Bangalore, on our Storage product line. Some of them had worked for our competitors, and it was fun to see the contrasts between product ranges that came out.
Sometimes it was “oh – our stuff couldn’t do that” or “we charged for that, you gave it away free”. On other topics, they had experience at an “Enterprise” level that I just don’t, and customers with Enterprise budgets. On at least one occasion my natural response, to a question like “why doesn’t this product do X?”, was something like “sure, it’s possible, but why would you want to, here? It would be hard to manage, and cost too much”.
Technological advances have a way of turning last decade’s impossible in to last year’s unworkable, then in to this year’s impractical, and next year’s you-should-be-doing-it-already. Without going in to too much detail here, one of the topics of contention in our discussions was that of “storage virtualization”. I have a certain take on it, from my work on our mid- and low-end product ranges, that is focused on the nuts-and-bolts of how getting spinning metal platters to do what the customer needs. The first difference between a standard PC and a Server is the use of multiple disks, how they need to be arranged to work, quickly and reliably, without costing the earth. Beyond that you get in to dedicated external Storage systems, which is where I make my living.
There’s another level beyond that, an additional layer of virtualization that hides the complexity from the customer and allows them to treat storage as a utility (like electricity or water), but that comes at huge financial cost, without actually removing the underlying complexity. I would use a vehicle analogy; modern cars, with all their engine management and monitoring, can run reliably for longer than old cars can. That is fine when they’re working, and a real problem when they go wrong, since the driver doesn’t have the knowledge or tools to fix it any more. There are people who do, but they’re expensive, even with service contracts, and that leaves you at the mercy of the service personnel.
Worse, this dumbing-down of the end users leaves them unable to make fully-informed decisions about their next purchase. My company has Storage customers who are the technological siblings of SUV-driving Yuppies in cities; they fall for the sales pitch and a misplaced conception of safety. Others buy the equivalent of a VW Golf from us – great for its purpose, but these customers then turn round and complain to us that it doesn’t perform as the sales pitch said, after they pour diesel in the petrol tank, and load up the back seats with a ton of cement. They have a service contract, so we must fix it. Right?
Anyway, to get back to the main point: I’m going to have a go at writing up my ideas on Storage, starting with the most basic abstract view of what it is and isn’t. There are so many storage types, real and virtualized, and while it can be confusing, you can cut through it all with a solid grounding in the fundamentals. What I write will appear under the “storage” heading under Pages, not as blog entries in the usual way, though I will refer to them at times. It won’t be quick!
Not much happened over the weekend, though on Sunday I undertook the kind of walk I should do more often. If you know how you’re getting home, that means you can walk until you drop, or at least get bored. I ended up at the end of Dun Loaghaire’s South Pier, about six miles down the road, and did most of the way back, barring a short segment by DART train. The sun was out at least half the time, so I was complemented on my “nice colour” by a lady friend today.
The b3ta bodges I’ve been doing have sprouted a major sub-category of their own, which I call the planemashes, and these are now on a web page of their own, The latest, #8 from last Sunday, was something of a solution in search of a problem, and the results may be slightly offensive to the humour-impaired. Not that that’s ever stopped me before.
I don’t really want to say more about the crash self-training I’m undertaking on our NAS product family, except to say: setting up and using a system of this type is like being on a package holiday where your schedule is laid out in advance, and any deviation means the bus can leave without you. Great for amateurs, but, to stretch the analogy a bit, my colleagues and I are already well capable of taking care of ourselves in the computing world.
We can lasso a passing metaphorical mustang and ride it off into the sunset if we have to, yet here we are taking the bus. It will still get us there, but it takes longer and the only fresh scenery you see is when the bus goes off the road, and you’re staggering through a forest going “where the hell are we?” This is why I may be heading to Scotland (not London, as was I previously told), to a bus that’s gone off the road because the driver crazily thought it had four-wheel drive. We need to tow the bus back on to the road, feed the driver some Valium, and send him off on the straight and narrow again.
Why? Because customers have asked for this “simple” system, and the customer is always right, even when they’re wrong. Isn’t it funny how, in my years of using my employer’s products before I joined the company, I never found myself having to call in for support, or expected to be spoon-fed in this fashion? To put it another way: having a fixed route from A to B can work fine, until the road is blocked. That’s when we find out who will cut across country, and who will just sit in the traffic jam.
Holy Moley, Batman! It’s already Friday, and the end of February. It’s been a busy week, compounded by another 2-day session of the Enterprise Backup Systems training course I run here. The other guy who was sharing duties here has pulled out, so I’ll be doing them all on my own from now on, unless we can find some other sucker to take it on.
The material is outdated, and badly structured, so that the whole first day is “Death by PowerPoint”, with a few of the students tempted to jump out the window for some relief. I’m tempted to follow them, so I try to liven thing up with jokes. Last time I told the story of Murphy’s Law; this time I had the students imagine they were trying to degauss (wipe) a backup tape by putting it on top of a loudspeaker. I asked for their opinions on which music would be best at the job; I lean towards Industrial music with heavy clipped square wave content in the bass range.
- It takes more power to produce a bass note at a given volume than treble – a woofer needs hundreds of watts, but a tiny piezo tweeter using a fraction of a watt can deafen you.
- OK, square waves aren’t as “dense” in the frequency spectrum as sawtooth waves, but square waves are easy to produce: just overload the signal until it clips. Do this to a guitar and you get “fuzz”, do it to a bass and you get Nine Inch Nails.
There were several Germans on this course, so Techno was quite a popular suggestion. Kraftwerk? Nein, not dense enough. I’m thinking the kind of bangin’ Techno they pump out at the Love Parade every summer in Berlin, at the kind of SPL found there too. The magnetic leakage from the speakers must be frightening to anyone with a pacemaker, and it wouldn’t be good to any magnetic media such as tapes or disks.
Today I booked my main summer holiday – some old school friends are renting a cottage in the Highlands in July, so we’ll all get together and catch up. I may take a Risk game with me, we got a lot of mileage out of that before. The last time we were all together, in 1986 I think, we bought a lot of beer, had a barbecue, then started playing Risk at about 5PM. We were still fighting board battles at 2am, the beer was gone, so we raided the drinks cabinet, and by 5am I sat down to “rest my eyes” for a minute and woke up three hours later, still drunk.
I’ve never had that kind of drinking experience again, because everyone drinks so much so quickly now, I don’t even try to keep up. Back at that party in 1986, we drank steadily but slowly for over 12 hours, and it didn’t bother me. I avoided a hangover by not going to sleep again, after that three hour nap, until that night.
What a week, starting with the Enterprise Backup Solutions (EBS) training course I give. As before, the biggest problem is actually getting hardware for the students to use to build a backup solution. It calls for one setup per three people, and a typical EBS setup includes a server with Fibre Channel adaptors, a SAN Switch, a “Data Router” that translates data and commands between Fibre Channel and different SCSI cabling types (depending on what cards you have plugged in).
Then there’s the “library”, which in generic terms is a bunch of tape drives inside a housing with tape storage slots and one or more “robot” mechanisms that shift tapes around between slots and drives. A library robot operates in almost the way you imagine, looking and sounding like something from Blue Peter or the Discovery Channel.
Libraries vary in size, from a rack-mounted box 2U (3.5 inches) in height, to a free-standing cabinet that takes up half the room, with a capacity measured in tens of terabytes. In the latter case the robot will not work if the door is open, for safety reasons. (It could rip your arm off at the shoulder.) A library is a pretty dumb system, and it relies on software running on the server to control the robot(s) and manage the tapes.
The robot part of the system doesn’t go wrong much, but we have far more problems with the SCSI communications side, with each 68-pin cable practically a fault waiting to happen. It’s a prime candidate for replacement with new technologies such as Serial Attached SCSI (SAS), which has the potential to be more reliable.
This is a field I need to keep one eye on, and writing a little piece like this helps me get my thoughts in order. In case you’re wondering why I’m doing this…
Still no laptop. The store says “this week”, and they had better mean it.The training course I gave is over, without major incident. The main problem is that I was seeing things on the slides that
- I wasn’t aware of, or
- I don’t agree with.
A typical example is the order in which you start up a SAN system. Since the Switches may be used for other SAN functions, like disk storage on critical servers, you don’t go switching them on and off willy-nilly. It’s fair to say that they are the SAN, or at least its backbone. One of the slides advocated taking them down as part of the process of starting up the backup systems. Well, if you deliberately crash a critical server by disconnecting it from its storage, you should keep a backup copy of your CV or resumé at home. On paper, since you can’t be trusted to operate a computer.