For IT architects, one of the most important non-functional requirements to determine is the availability needs of a system’s users. It’s often expensive and risky adding availability features to an already deployed solution, so getting it right first time is important. In current times however, we’re being asked to regularly …
I could be wrong, but I dont think even the "big boy clouds"™ quote 100% availability?
Certainly I remember most have had one outage or another over the last year or so. Doesnt take much to knock those numbers into the 90s.
Re: 100% availability?
Lots of service providers will give a contracted 100% SLA, but the penalty for it being breached is sometimes only a refund of the pro-rated service cost e.g. your £1m/yr cloud service is down an hour so you get £114.15 back.
Re: 100% availability?
This already exeists for numerous customers does it not.....online banking, ATMs, hospital systems, tax systems. If the front end is running 24 hours then why not the backend, auditing and reporting mechanisms?
Re: 100% availability?
This already exeists for numerous customers does it not.....online banking, ATMs, hospital systems, tax systems
What stone have you been living under? One word: RBS.
Re: 100% availability?
Can't tell you about the others, but online banking and ATMs certainly don't come with a 100% availability. There are maintenance windows and just general failure from time to time.
Users may *expect* 100% but the reality is very different. Highly available services (the stuff you don't want failing, ever) is only usually rated to what was previously the holy grail of "five 9's" (or 99.999%), which equates to about 5 minutes outage per year as I recall.
There's also the question of whether availability even means anything. It's a very crude tool to measure outage impact. A 30 minute outage on the ATM network is more costly to a bank at 5pm than 4am, but availaiblity stats will show the same figure.
Re: 100% availability?
*shrugs* The mystical cloud is the current in-thing among the masses. Got some sort of IT problem? Doesn't matter what it is, it can probably be made better by invoking the cloud. See attached image: http://img3.etsystatic.com/000/0/5740779/il_fullxfull.110833487.jpg . And attached YouTube: https://www.youtube.com/watch?v=ApLLTV04Gbc .
Re: 100% availability?
I prefer this one-- http://imgs.xkcd.com/comics/the_cloud.png
Gavin Payne is a Microsoft Certified Architect who scopes, designs, implements and migrates mission critical data platforms
Oxymoron or lintard?
I find the whole "I'm smart because the OS I've chosen to use makes me smarter than people who use other OSes that I don't approve of" thing really rather tedious. Can you say confirmation bias?
RE: Confirmation bias
You'd better get used to that around here...
The comment was badly put, but when someone says their ipad is more reliable than their kitchen appliances, I begin to roll my eyes and wonder about the speakers credentials. I haven't had a service failure from my tumble dryer since purchase, over 10 years ago. Given the relative youth of the ipad, I begin to think that the windows expert must be rather young and/or inexperienced. Then I generalise the experience, including all the other windows support staff I've met and come up with a correlation (not causation)-inspired generalisation that windows people are not really enterprise-ready. In contrast, *nix people I've met tend to be to work on enterprise and carrier-class systems, but not even carrier-class systems run for that long without needing attention.
It isn't necessarily a reflection on the author himself who may have just had a bad-example day, or been unlucky with his white goods, but when you speak publicly, your speech reflects on "your kind" of people. Identifying himself as a windows person, associates him with other windows people and other windows people with him. Tarred with the same brush, rightly or wrongly. It isn't fair, but generalisation is how we deal large numbers of things.
Since the average bash n00b can script an average MSCE out of a job, I do actually agree that the learning curve of a preferred tool set might actually have some correlation to a those tool's user's intelligence. However, this simple measure should not be taken for scientific proof that admins who prefer MS products are brain dead. Naturally I would suggest seeing a doctor. After all, you'd believe him whether or not what he is saying actually true because he wears a white lab coat.
Whilst I agree that business users now expect services to be running 24*7 or more likely just when *they* are awake...the current trend in news reporting of network outages means that people are generally more aware that Twitter, Amazon, Blackberry and mobile networks are not 100% reliable and so that expectation has slightly lessened.
Also, even those Apple devices (iPads, iPhones) need software updates and so the explanation that the system needs an update is actually understood and appreciated by users rather than them demanding a non-critical system stay available whilst updating.
What's the problem?
It all depends on the end users making a business case - if they believe (and can show) that 24/7 availability of any system is necessary then it can be provided, at a cost. It is up to the bean-counters to decide whether the cost of having support available 24/7 plus the necessary resilient hardware and software environment is justified by the business need. If you are talking about a global online sales system, where the customer can give you money, then the cost of losing that for 24 hours is high. The real cost of failing to allow some obsessive marketing drone to check the latest usage stats before they go to bed at 2am is probably rather less. Maybe there is a place for some negotiation here.
This is nothing fundamentally new - the same principles applied when developing and supporting mainframe systems decades ago. What will it cost to provide a particular service level? How much is having (or not having) that service level worth to the business?
It's a simple balance...
...availability vs. cost.
There have been highly available and even un-interruptible services around, but you have to pay for them, and they don't come cheap. Nor do the staff to set them up and run them.
Microsoft probably have an HA offering now, but I expect that even they will charge a premium.
Segregating the function, so that you can put your information distribution systems on simple, small, cheap, and redundant servers in front of your actual service machines can help with the appearance of a service being available (as well as increasing the security), but if you truly want high availability, it's going to cost.
"My iPad is more robust than most of the appliances in my kitchen never mind an enterprise data centre"
I got that far before it became apparent this was written by someone who knows absolutely nothing about the contents of an enterprise datacentre, or kitchen appliances.
When the author comes back in 25 years and tells me that his 2012 iPad is still working and has never broken down, then I'll believe that statement!
Heck, my mother was still using her Sunbeam electric hand mixer in 2010, which she got as a wedding present 45
I've certainly still got kitchen appliances in daily use that I bought when I got my first house, back on 1990... And they'´ve been bumped around, moved across continents, dropped several times, cleaned with abbrasive chemical detergents... But they are still going strong.
I made it as far as "For IT departments with staff that grew up with 9 to 5 working hours"...
Or indeed iThingies
I gather the author means 'physically robust', because even though some kitchen appliances may contain a microcontroller with its associated firmware, the most common cooking utensils still do not.
Now go and accidentally sit on a rackmount server, on an iPad and on an egg whisk. Two of these will break and be out of order for a while.
Not So !
In September, 1937
I bought my wife a new electric iron for eight and sixpence
She's still using it everyday and it's never needed repair
The Bonzo Dog Band, Rockaliser Baby
"Now go and accidentally sit on a rackmount server, on an iPad and on an egg whisk. Two of these will break and be out of order for a while."
Your argument here is not exactly generic or all encompassing and really only works because you chose one of the most fragile of all kitchen utensils. Try "accidentally it on a rackmount server, on an iPad and on a rolling pin", Two of these will keep working without a hitch and it won't be the iPad.
In most cases the same applies for: wooden spook, knife, cooker, dishwasher, fridge, need I go on?
In terms of operating robustness I also disagree with the author and agree with many of the posters. My wife has a mixer that was passed on to her from her grandmother. It's built like a tank and will still be going long after any of the modern electronics items kicking around our house!
"Now go an accidentally sit on your own fingers while trying to type an El Reg post"... as I can testify from experience my typing breaks down pretty quickly! Wooden spook anyone?
RE: "My iPad is more robust than most of the appliances....
Well, I doubt it is as robust as the Kenmore washing machine I recently replaced, after 40 years of service.
I seriously doubt that your iPad would even last 10!!!!
What a twat.
I'm with you Gavin
My iPad 2's been working away happily yet I've been through a microwave, two kettles and a toaster - all of similar age.
Can't speak for your average datacenter but I imagine, where kitchen appliances are simple, datacenters are complicated enough to be more likely to suffer a component failure of some description than my iPad.
I admit it has a case and after a certain figure the height becomes irrelevant but it does still work. Even with a case I imagine my toaster would either break or try to kill me.
Re: RE: "My iPad is more robust than most of the appliances....
"What a twat. "
Now hang on, that's a bit unnecessary. A twit maybe...
"how do we learn to trust a face-less cloud service provider"
Why should a cloud provider be treated any differently to any other supplier? If the Architect is doing his job properly, the service will be properly scaled and agreed, SLAs in place, and penalties for failure.
The cloud is only face-less if you design it that way.
Another Microsoft advertorial?
>: the answer like it seems to be for everyone but the richest organisations will almost certainly be the cloud.
Just like when then only solution you have is a hammer, so every problem looks like a nail?
The correct solution is a business continuity plan, with properly performed risk analysis, to determine what your company needs, what the biggest risks to that are, and how likely those risks are to come to pass.
Based on that you can work out how much it costs to provide the protection you need, which of course will not be the same for each user and each department. I suggest that instead of simply waving your hands in the air and claiming that "our cloud" will solve everything for everyone, you take a look at how the professionals do it. Start with the UK-focused Business Continuity Institute (http://www.thebci.org/) or perhaps the more US-centric Disaster Recovery Institute (https://www.drii.org/) and their various publications and conferences.
Re: Another Microsoft advertorial?
"The correct solution is a business continuity plan"
A couple of months ago there was a power cut in our city centre - only lasted about five minutes. Almost all the shops closed crying "the tills won't work, we can't sell anything!!!" At the Post Office they were recording transactions in a paper notebook. The first transaction was taking the notebook out of stock.
Cost vs advantages
Pretty sure every IT department could offer near 100% uptime, with worldwide availability on a multitude of devices if they had enough staff and funding.
However, most don't. So, instead, they have to balance the perceived advantages of 24/7 access on consumer devices against their limited budgets. The reason Facebook can be always online? They have millions of servers and thousands of staff constantly adjusting, tweaking, upgrading, replacing and improving their systems. Most businesses question why you need a new server let alone a new data centre.
The consumerisation of business IT has made business IT professionals look slow or restrictive but they have a good reason to be - security. That iPad having access to all your financial details is a security risk. It might be fine for a consumer to have that, but a Fortune 500 company? Hmm...
Re: Cost vs advantages
Right. This stuff isn't magic no matter how mysterious it might appear to some. You wouldn't expect the phones to be answered 24x7 or the manufacturing lines to go three shifts without an investment in staff or gear, and IT isn't much different.
If it sounds cool and you think you want to be Always On! then call your staff and Get A Bid! because it shouldn't be free. There are costs to this that are being paid by those "consumer grade" companies like Facebook, and they are paying them because it's their bread and butter core business. Before your widget company makes a similar choice they need to decide how core this is, how global they are, and what IT is worth at 3:00am. "Just because you can doesn't mean you need to."
Re: "They have millions of servers and thousands of staff constantly"...
retweaking your privacy settings.
Oops, wrong thread.
Yet, with all those support people, Facebook is still not online 100% of the time, there have been failures.
Makes you step back and ponder that famous "cloud" wonderworld where everyone and his dog is offering to harbor your data.
Seems like it actually takes competence to maintain a cloud operation. Does every cloud provider have that competence ?
My iPad is more robust than most of the appliances in my kitchen
I doubt that. Most, if not all, of the appliances in my kitchen will still work after I deliver a swift kick from my size 9 boots. How well will your iPad fare? I'm available if you want to find out experimentally.
Re: My iPad is more robust than most of the appliances in my kitchen
How available? 100%?
I once had my motorcycle engine block on a cooker; I want to try that one on his iPad
Re: My iPad is more robust than most of the appliances in my kitchen
What exactly is the definition of "robust" here?
Ability to survive being dunked in the sink, dropped on the floor, banged on the worktop to remove crumbs, scrubbed clean with wire-wool and cream cleaner?
Re: re: boots
Didn't you have the block inside the oven?
Same on you. That's how you change the main bearings on most old British bikes.
My ex used to moan like mad when i used her 'spotless' oven for main bearing changing. Sadly my 1969 Trident Crankcase won't fit in my current oven.
Same discussion when everybody had Windows PC's
A lot of downtime of devices, appliances, networks and clouds alike will not go away simply because human beings will not go away. The tablet only seems reliable because it's mass produced, with very predictable behaviour to test during development and production. We can say landlines or TV are reasonably reliable because of similar reasons. But IT is in the real world a more complex beast especially when customized to the needs of a business or customer. If you install another OS on the tablet or heavily tweak the thing, I can assure you it won't behave that predictable any more. See also various form of hardware and software certification like they need to define reliability in high-end industries or just any standardized configuration management.
Indeed some customers will expect from a real dynamic or complex system the same as from their iPad's but that's not a new problem. When everybody had a Windows PC, people thought all automation, installing and management on every scale in every business was just as easy as clicking on a wizard. But the architects survived that and they will survive the personal device bias.
iPad vs Fryingpan.
iPad can have the first swing.
Let the games commence!
Re: A testing!
Having been hit in the face by an airborne iPad before (don't ask!) I'd rather have the egg whisk from earlier please...
Paris because she's used to foreign objects in her face.
The fact that staff want something
doesn't make it a legitimate business requirement.
I'd like a high floor office in the Shard and a private corporate jet. They'd undoubtedly boost my productivity a few percent but for some reason the logistics team don't feel compelled to provide them for me within their budget.
The fact that I can visit the Shard observation deck or buy flights online from a variety of vendors has yet to move them to meet my expectations.
The catering staff won't even bring a muay thai to my back garden when its sunny!
How dare business segments make decisions based on strategic objectives rather than to validate individual staff members.
Re: The fact that staff want something
I'd like a laptop that works, but apparently that's not in scope. Honestly, the keyboard makes my type like Officer Crabtree from Allo Allo some days.
Empowering cutting-edge infrastructures
Personally speaking, I've found that by simply visualising virtual models and embracing open-source interfaces for user-centric channels it's been easy to deliver end-to-end user experiences that redefine world-class collaborative availability.
just my 2k
Re: Empowering cutting-edge infrastructures
Oh dear, oh dear, oh dear...
"Gavin Payne is a Microsoft Certified Architect who scopes, designs, implements and migrates mission critical data platforms. You can hear him share his knowledge and experiences at the "Iasa UK Architect Summit - Enabling Disruptive Innovation"
Who is it you work for Gavin? Because if this written perspective is the sum total of all of your "knowledge and experiences" then it looks to me that you've not got much of either and should move on. To put your head above the parapet and scribe this drivel - and for El Reg to publish it is embarassing and patronising to the professional IT populi.
If you came into my offices, sat with my Architects and suggested that we should... errr... hang on, what was it you said we should be doing with our mission critical data systems... oh yes... "what we need to be doing is blurring a solution’s non-functional availability and reliability requirements"? Yes? Presumably by trying to sell a Microsoft cloud solution is it??? Well quite frankly you'd be laughed out of the offices, all the way down the boulevard, and out into the deep ocean.
Good luck with your conference.
Agree but it means slickly MANAGING EXPECTATIONS....
Anything from an MS architect always needs to be scrutinized for possible sales angles etc.
However, sometimes there are in-house EXPECTATIONS, i.e. requests from senior management-- the types who get their tech briefings from in-flight magazines...
Tech media frequently hypes the-emperors-new-clothes, presenting a host of new platform options that appear RISK FREE, with little regard to security or practical implementation. Sometimes you have to take executives aside and DELICATELY suggest that maybe they need to Curb their Enthusiasm a little...
A past example comes to mind. Senior executives managing billion dollar businesses wanted to use GoToMyPC and similar services to get access to their Desktops in order to see their spreadsheets, access mission critical in-house apps, and use Bloomberg etc. Corporate IT had long banned these. But I was being paid by the business not the tech side, and they wanted me to game-the-system.
It took a Herculean effort to point out some of the risks i.e. that their proprietary business models could be exposed etc... They hadn't thought about that, only about the added convenience of logging in at the weekend using an unknown computer belonging to their most recent conquest...
Reality check for lotuseaters
Full 24x7x365 at 99.5% means 43+ HOURS per YEAR with NO service .
They want/expect 99.9995, something like 40 mins/year. Electricity or running water like.
Taken us decades to get that degree of availability.
And how much investment??
Where is the money? where the techies?
Architects are fine, BUT then you have to actually build the thing, and then the TEAM of engineers,etc. get to do it.
The architects normally say it looked great on paper/scale model.
And then came PowerPoint.
And it looks beautiful.
But nobody listens to the bricklayers.
Re: Reality check for lotuseaters
Totally agree, but go back and read again. It looks like he realizes that only offering 99.5% looks bad so we should offer 24x7 and a good RTO rather than a specific uptime number. Which is a totally bogus faux-metric that the cloud vendors dreamed up. I'd rather get a solid schedule with an identified service window than some gob "blurring a solution’s non-functional availability and reliability requirements" so that when the system inevitably goes down I can steer that away from business hours to a maintenance period after midnight.
Gavin, if you showed up in my office offering this pablum you'd face a short period of rude questioning followed a rather abrupt end to the meeting.
Re: Reality check for lotuseaters
You lost me at 24x7x365
Perhpas you were being metaphoric.
There are only approximately 52+5/208 weeks in a year, or does it mean 7 years. There are leap years.
- Fee fie Firefox: Mozilla's lawyers probe Dell over browser install charge
- 20 Freescale staff on vanished Malaysia Airlines flight MH370
- Did Apple's iOS literally make you SICK? Try swallowing version 7.1
- Neil Young touts MP3 player that's no Piece of Crap
- Review Distro diaspora: Four flavours of Ubuntu unpacked