Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Microsoft/Danger lose Sidekick user data in server failure

forums.t-mobile.com

Post reply on HN
Print view View original

61–70 of 70 posts

Re: Microsoft/Danger lose Sidekick user data in server failure

#61

Post by viraptor »

johnnybgoode wrote: post #51
viraptor wrote: post #26

Earlier quoted context omitted.

It doesn't have to be the case of no backups. Many different things can happen... They could have a faulty backup design. They could have lost the backups. They could have a fault in their backup verification / checking procedure. They could miss some part of the data that wasn't crucial before, but after a couple of updates, they cannot restore contacts without it. Etc. There are many ways this could have happened -…

As far as I'm concerned, if you can't recover your data, you don't really have a backup. That covers every one of your hypotheticals.

But you cannot be sure you can recover your data (you can be sure in theory only).

In practice, you don't know if you can recover your data unless you migrate the whole company to the backup copy and ask them to check every functionality they use -- every time you make a copy (and rely on them doing 100% coverage). The best you can have during the normal operation is a high probability that what you get is a proper backup, because that would be the case the last time you did a full test (if you ever did). Datasets change, ways of interfacing with data change, etc. Testing is not perfect either, because your tests can be broken or can be reporting false positives.

So yeah... noone can be sure they can recover the data. Does that mean that noone really has a backup?

Re: Microsoft/Danger lose Sidekick user data in server failure

#62

Post by chrischen »

chrischen wrote: post #40
jonursenbach wrote: post #27

Earlier quoted context omitted.

I don't think this is really showcasing the dangers of the cloud, but rather the dangers of companies not knowing what the fuck they're doing.

Yes and these companies who don't know what the fuck they are doing make up the clouds we use everyday, so the cloud ends up being a hazy fog of data uncertainty instead.

What I meant was exactly what btn said. Fucked up companies are behind the clouds, so this does show the dangers of the cloud.

Re: Microsoft/Danger lose Sidekick user data in server failure

#63

Post by johnnybgoode »

viraptor wrote: post #61
johnnybgoode wrote: post #51

Earlier quoted context omitted.

As far as I'm concerned, if you can't recover your data, you don't really have a backup. That covers every one of your hypotheticals.

But you cannot be sure you can recover your data (you can be sure in theory only). In practice, you don't know if you can recover your data unless you migrate the whole company to the backup copy and ask them to check every functionality they use -- every time you make a copy (and rely on them doing 100% coverage). The best you can have during the normal operation is a high probability that what you get is a proper b…

Not to mention the planet could explode suddenly, so yeah, no one can be literally 100% sure that they can recover their data.

> Does that mean that noone really has a backup?

No, it means in some abstract sense no one can be 100% sure they have a proper backup. But there are some fairly obvious ways in which you can get that number pretty close to 100%.

Re: Microsoft/Danger lose Sidekick user data in server failure

#64

Post by chronomex »

allenbrunson wrote: post #53
white_eskimo wrote: post #52

Earlier quoted context omitted.

Server side re-rendering of websites?? Sounds like wide-spread support for that would be lacking... I'm sure as the Sidekick gained traction, more web sites adopted its special rendering requirements as you mention, but in the beginning, things must have been rough! Disclosure - I've never owned a Sidekick (and know little about them, other than they are apart of Microsoft's 3 screens and a cloud strategy, yet they d…

i'm pretty sure he's talking about servers run by danger that re-render other websites. those other websites don't know or care that they are being presented in such a fashion.

Opera does this for its free Java MIDlet browser (Opera Mini). They also resize and recompress images into JPEG.

Re: Microsoft/Danger lose Sidekick user data in server failure

#65

Post by paul9290 »

Elepsis wrote: post #50
MattCutts wrote: post #36

Earlier quoted context omitted.

Or the dangers of companies not allowing you to easily export/backup your data.

While this is true in principle, this is a bit of a Google talking point that has more PR cachet than practical implications. Giving users the ability to make backups of their data certainly does not mean they will actually do so. How many Gmail users actually keep full backups of their email (using POP or IMAP or whatever)? And if Gmail had a case of massive data loss tomorrow, do you really think saying "well, you…

I back up my Gmail by forwarding it to other providers (Yahoo, Comcast). If Gmail fails and loses all my data, it's in my comcast and yahoo account. If they both fail then ok Im screwed, but the Internet I hope would have to melt down if all three of these systems failed simultaneously.

Re: Microsoft/Danger lose Sidekick user data in server failure

#66

Post by joe_the_user »

Elepsis wrote: post #50
MattCutts wrote: post #36

Earlier quoted context omitted.

Or the dangers of companies not allowing you to easily export/backup your data.

While this is true in principle, this is a bit of a Google talking point that has more PR cachet than practical implications. Giving users the ability to make backups of their data certainly does not mean they will actually do so. How many Gmail users actually keep full backups of their email (using POP or IMAP or whatever)? And if Gmail had a case of massive data loss tomorrow, do you really think saying "well, you…

Uh,

"Data Liberation", as in educated users, is the solution.

You back up your data because no cares as much about your data as you do (repeat...). It's like basic hygene. In the future, few people will be actual computer experts but everyone will know a few things. Backing up data will be one of them. A large enough density of people who understand this will mean those who will lack excuses - see main story!

Re: Microsoft/Danger lose Sidekick user data in server failure

#67

Post by rbanffy »

wrinklz wrote: post #55
MattCutts wrote: post #36

Earlier quoted context omitted.

Or the dangers of companies not allowing you to easily export/backup your data.

That doesn't apply to Danger/Sidekick. I had one 4years ago and synced regularly on PC and Mac.

One nice thing of having your data on the server instead of on-device is that any syncing can happen between the server and your desktop, without even the need for USB cables.

Re: Microsoft/Danger lose Sidekick user data in server failure

#68

Post by dtf »

rbanffy wrote: post #59

I smell a migration of their infrastructure to Windows and a nice "Get the Facts" page about this disaster, how it could be blamed on non-Windows technology and how everything was saved (or, at least, made safe) by the deployment of Windows Server 2008.

Word on the street is that it was a botched SAN upgrade by a Hitachi tech. Of course, the fact that the Danger admins hadn't backed up the SAN before the upgrade made that fatal. Not even an offsite backup... crazy.

Re: Microsoft/Danger lose Sidekick user data in server failure

#69

Post by rbanffy »

eli wrote: post #41
lallysingh wrote: post #2

This is really, really unacceptable. I don't know how you screw something up this badly. HINT: http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&#3... Additionally, I suspect we're going to see these customers really hesitate on future cloud-based services.

I'm guessing they have a lot more data than is practical to store on a tape.

You do it by having redundant storage.

I am admittedly not a good sysadmin, but I keep an eye on my own "pirate" server (that is, outside the realm of the corporate network) and I have already ordered additional storage because Munin shows that, at present consumption rates, we will run out of storage in about 4 months, crossing the 70% mark in three. I know they will take at about two months for corpotate IT to spend the US$ 100 it costs for a pair of disk-drives and that gives me a month to set-up a soft RAID before I cross the 70% mark.

It's completely unacceptable not to have the storage mirrored and load-balanced on a second SAN. And mind you many SAN vendors even bundle software that makes it easy to do. Even if you don't have the budget to buy another SAN the same size, you can mirror the data rather easily with a bunch of servers, each with a bunch of SATA cards and a bunch of low-budget drives. Redundancy is the key here - if you have a drive that's 10x cheaper than the same storage in the SAN, you can replicate in a RAID1 with 10 drives. If each drive is 30% as reliable as the high-grade disks that go into the SAN, you are well on the winning side.

And I seriously doubt there is that much data. How much? A couple dozen terabytes? That's fits easily under my desk. And quite probably in my discretionary budget.

Re: Microsoft/Danger lose Sidekick user data in server failure

#70

Post by rbanffy »

dtf wrote: post #68
rbanffy wrote: post #59

I smell a migration of their infrastructure to Windows and a nice "Get the Facts" page about this disaster, how it could be blamed on non-Windows technology and how everything was saved (or, at least, made safe) by the deployment of Windows Server 2008.

Word on the street is that it was a botched SAN upgrade by a Hitachi tech. Of course, the fact that the Danger admins hadn't backed up the SAN before the upgrade made that fatal. Not even an offsite backup... crazy.

And that would be one of the bullet-points in the page: how hard it is to properly back-up stuff in *BSD as compared to how easy and foolproof it is under Windows 2008 Server.

In fact, I can almost smell a migration directly to Windows 7 Server or whatever the name they decide to use for that.

Post reply on HN