On-Line Data Storage
The BBC reports that Microsoft recovers Sidekick data
Microsoft says it has now recovered the personal data lost when its Sidekick servers suffered an outage on 13 October.
The online service backs up contacts, photos, calendar appointments and other personal data stored on the mobile phones of Sidekick T Mobile customers.
Microsoft Corporate Vice President Roz Ho says that all data will be restored, beginning with personal contacts.
She believes that only a minority of Sidekick users are still affected.
“The outage was caused by a system failure that created data loss in the core database and the back up,” she wrote in an open letter to customers.
My guess is that they rebuilt the data base, which is often successful if the structure was designed in accordance with industry standards. As the original was not designed by Microsoft it is possible that most people will be able to access their data. A minority of users is one less than half, so any number below a half million is a minority, but it is probably a group in the thousands, and not tens of thousands if the rebuild was successful. My guess is based on the amount of time involved. If they had found a valid backup, they would have been running in a day.
The BBC also had this piece, Call for rethink on data storage
The rules on the movement of electronic data around the world need an overhaul, according to a Microsoft executive.
Many technology companies, including Microsoft, IBM and Yahoo, use vast data centres to run services over the internet and store data such as e-mail.
However, the rules for data storage can vary widely from country to country.
One might insist that data is kept for a year while another might want it destroyed in six months, said Microsoft senior vice-president Brad Smith.
My solution would be distributed data centers that operated according to local laws rather than massive centralized systems that might be less expensive, but are more vulnerable to problems. I’m sure that T-Mobile would have preferred a distributed system, rather than the single center approach.
I’ve worked in both environments, and the centralized system is cheaper if nothing goes wrong. There are fewer people and less equipment needed for the same volume of traffic, but if there is a “weather event”, an accident, or a deliberate attack, you are better off with a distributed system. That is, after all, why the Internet was designed the way it is – to route around problems.
11 comments
Yeah… ‘What could go wrong?’ 🙄
Of course, anyone with a few working brain cells knows that if the name ‘Micro$oft’ is attached anywhere in any way, the chances are better than even that it will go horribly wrong. You’d think after so many years of proof positive, people would run a mile rather than use any M$ service etc. 😳
People are stupid and stupidity is it’s own reward. 😐
Backup? Wat dat? we don’t need no stinkin backups! We make money!! Lot’s of money! We don’t spend nuttin on no backup systems! Teh stoopids will keep payin even if it breaks! Ha!
Yup. M$ have your number. They spend more on profilers, lawyers and marketing guru’s than they ever will on product safety and reliability. Just be really happy M$ haven’t yet got into making Military systems, or medical systems… oh, wait… Win Embedded CE! 😮
“Many technology companies, including Microsoft, IBM and Yahoo, use vast data centres to run services over the internet and store data such as e-mail.” – BBC
That’s almost correct. Clearly, Microsoft uses half-vast data centers.
.-= last blog ..Nothing To Say, And Not Saying It =-.
The joys of dealing with a monopoly that keeps buying up small, innovative companies and milking their products for profits, while destroying them in the process.
That was definitely worth a GRRROOaaannn, Steve.
Swats Steve with a rubber chicken….
Linden Labs’ Second Life deals with the stored data of millions of accounts (as opposed to customers as each customer may have more than one account). They use a distributed system spread across the USA. I believe, and I could be wrong, that they also have server farms off-shore.
Not only are the computers used to run the virtual world distributed, the simulators which service each region, estate or island, and the asset servers that hold the data associated with each account are also spread out. Back-ups of region simulators occur at least every couple of days.
A weather event or other mishap that disables a server farm might make the whole system slow and wonky until the problem is solved, but it would take a major, nation-wide disaster to bring the grid down completely. If such a disaster were to happen, I doubt very much that people would be worrying about playing Second Life.
One might wonder why such care is lavished on what appears to be nothing more than a MMORPG, but Second Life has its own economy probably equivalent to a small European country (and growing) and has become a corporate work tool, and educational aide as well..
A far cry from a virtual address book…
Actually gaming is the motivator for most IT advances these days, not just multimedia, because of the massive resource use. Multiplayer games are an ultimate test of real-time, interactive systems, so distributed systems and fault tolerance would be vital to success. If Second Life or WoW had been down as long as Sidekick the reaction wouldn’t have been resigned indignation.
The other important point is that the people at the top still understand their product and their customers, as opposed to most of the major IT companies which are “led” by MBAs who need people to deal with their e-mail.
The NetApp servers Danger used automatically snapshot once per hour. My theory is that they went back to the last hour’s snapshot before the crash and rebuilt the database from that. This also explains the amount of time it took — they would have to copy the data out of that (read-only) snapshot to someplace current, then run a rebuild program/consistency checker since it was an open database at the time it was snapshot. It takes a *long* time to copy a terabyte or two of data, even if you have 10gb network backbone and fast RAID arrays…
And why they were saying that the data was unrecoverable? Well, because Microsoft fired or drove off anybody who knew anything about the NetApp servers that Danger was using, doh! Windows Server doesn’t snapshot, so of *course* they wouldn’t know about that.
.-= last blog ..Deadly weapons =-.
They operate a huge Oracle database system and have to hire an outside contractor to do an update!?
They don’t seem to have a real sysadmin for their server farm, or there would be backups and the update would have been tested before going live.
It almost sounds like they are using day labor for their server farm, or temps at best. It certainly doesn’t sound like a professional organization. I don’t expect anything like an IBM mainframe corporate operation that I started out on, but it doesn’t sound like even the mini-based university systems staffed primarily with students that I started working on as a sysadmin.
I just can’t understand how a major farm can be that slipshod and remain in business. I guess it is part of the arrogance of being a monopoly.
Hi y’all… I was checking out a blog (Fileslinger – about backups etc.) and there was a blog about this that I thought I’d post as an update. 🙂 Interesting…
Putting Your Data in Danger
Boy… Talk about asking for trouble! Sheesh.
It is a bit disheartening to know that this blog has a more in-depth back-up system in place than a major commercial system.
I hate to break it to the Register, but paper only works until there’s a fire or flood.
Ummm… Le Reg knows this. 😉 And so do *most* of Le Reg’s regular readers. 😉 LOL
Le Reg is kinda like ‘The Daily Show’ they have to wrap up the news in comedy so they can be uber critical of coy’s like M$ and not get sued. 😆 Besides.. we looooove satire! 😀
Oh, that one. Sorry, I forgot about them, as I’m not a regular reader of tech snark, I get too much of it in real life.