This is one of the biggest steps back for Cloud computing. Some people where concerned about putting all the information on third parties companies, but now the cloud computing haters will have something more to be feed of:

T-Mobile: Microsoft Lost All the Sidekick Backups (http://www.datacenterknowledge.com/archives/2009/10/10/t-mobile-microsoft-lost-all-the-sidekick-backups/)

"The incident is already being cited by some news outlets as a confidence-buster for cloud computing. This comes up with every major service outage – especially one involving Google or Amazon. The design of the Sidekick service and backup practices of Danger/T-Mobile will be closely scrutinized - assuming a detailed “how this happened” explanation is issued. As it certainly should"

Welcome private clouds !!!

This is for sure another cloud buster, but in a major scale.

Wow, pretty huge... :S

I wonder what's going to happen for all the people who lost their data...

Very unfortunate :(

Yikes. As a past Sidekick user myself, I can vouch for the fact that this is pretty disastrous for some of the current Sidekick users. :(

Before this, I was a huge Danger fan! Now I am not so sure.

-mike

This is for sure another cloud buster, but in a major scale.
I'm trying to understand how this is a "cloud buster" at all. The article is entirely unclear about why this will reduce confidence in the "cloud."

Was this data stored on Amazon EC2/S3, GoGrid, SliceHost, etc.? There's no indication I can see that it was.

If the inference is that Danger's and/or Microsoft's storage servers and backup for the Sidekick data qualifies as "cloud" storage (storage that no one other than MS/Danger can access)- rather than a corporation that simply blew it as far as properly supporting their mission-critical software/hardware goes - then well I suppose it qualifies as a failure of the so-called "cloud," even if the infrastructure is entirely private and I doubt it is designed to be provisioned as "cloud" storage capacity.

In any case, it seems only reasonable that if this - the Sidekick data, that is - is what is supposed to qualify as "cloud" storage then it appears anyone who runs a private file server is also apparently providing "cloud" storage even if they didn't know it.

The notice is up on www.sidekick.com

As for the article naming the cloud name, its not about the technology but the concept. As far as I concern cloud exists for years, it was called grid before, they just added the pay per consume model.

The cloud model it refers is where people and users trust their data 100% to the cloud/service provider whithout any kind of local backup or copy.

Also, most clouds aren't that sure. More then 60% of data disasters are caused by software corruption. So this is something that CAN happen in the cloud. Data is only safe if its on another server/hardware, a full copy and even better offsite.

Last time I asked Softlayer about their iSCSI and cloudlayer they said they dont DO offsite backups of it. Its runs on RAID disks (SAN boxes) they would not specify any kind of guarantee or SLA on data lose. They basically just said, the data is safe as its on redundant disks. And we all know one thing for sure. RAID is not backup !!! Not even if its on a SAN with 100 drives.

Its impossible to replicate all the data cloud. Most providers just protect the hardware side or have some kind of snapshots (that are useless if originals are gone). So I would advice ALWAYS to have a copy of data, even in the cloud.

I just mention this because some people go the Cloud approach because they think because their data is on multiple systems at the same time its safe and its impossible to lose it. They do this to avoid making backups and think the cloud is the more stable and secure system. This is what I have read from people using mozzo, amazon and SL clouds. On real life they do have downtimes, even more then people that don't run on the cloud. I just wonder how secure their data is...this is just one example.

Don´t get me wrong. I don't have anything about the cloud model. I actually embrace it. But I think there should be millons of cloud providers and private clouds, not just 2 or 3 companies where 90% of the world puts their data. I prefer the approach VmWare is going, to develop private clouds, where small companies and clients can build their own network.

The notice is up on www.sidekick.com

As for the article naming the cloud name, its not about the technology but the concept. As far as I concern cloud exists for years, it was called grid before, they just added the pay per consume model.

The cloud model it refers is where people and users trust their data 100% to the cloud/service provider whithout any kind of local backup or copy.


-- You mean just like 90+% of the people out there today in the shared, dedicated and VPS markets?


Also, most clouds aren't that sure. More then 60% of data disasters are caused by software corruption. So this is something that CAN happen in the cloud. Data is only safe if its on another server/hardware, a full copy and even better offsite.

-- Cloud based solutions allow you to take an application (collection of vm's performing an overall service). Make it into an automated template and easily replicate it across multiple data centers. Typically at a lower cost then it would take to do the same with colo or dedicated platforms.


Last time I asked Softlayer about their iSCSI and cloudlayer they said they dont DO offsite backups of it. Its runs on RAID disks (SAN boxes) they would not specify any kind of guarantee or SLA on data lose. They basically just said, the data is safe as its on redundant disks. And we all know one thing for sure. RAID is not backup !!! Not even if its on a SAN with 100 drives.


-- Yes the backups would be your responsibility unless it is a 100% managed solution or you purchased backup services from them or a 3rd party like R1Soft, Rsync.net etc. SL is also not using the same cloud platforms as everyone else. Others providers use systems which have underlying protections in place to replicate your data in case of a hardware failure within your storage environment.


Its impossible to replicate all the data cloud. Most providers just protect the hardware side or have some kind of snapshots (that are useless if originals are gone). So I would advice ALWAYS to have a copy of data, even in the cloud.


-- This is incorrect. You can replicate data in the cloud just as easily if not easier then you can today with dedicated, shared or vps environments. Cloud based solutions just make it easier as it will be template based and also allow you to easily snapshot your entire environment and migrate or copy it to a remote site as a backup. This of course can also then be automated to make the overall backup process easier and optionally use secure tunnels such as VPN, SSH to perform the tasks to make the overall process more secure then people today who use standard ftp based solutions.


I just mention this because some people go the Cloud approach because they think because their data is on multiple systems at the same time its safe and its impossible to lose it. They do this to avoid making backups and think the cloud is the more stable and secure system. This is what I have read from people using mozzo, amazon and SL clouds. On real life they do have downtimes, even more then people that don't run on the cloud. I just wonder how secure their data is...this is just one example.


-- Mosso, AWS/S3 and CloudLayer are not the only cloud providers. Please do more research. You are missing a good 25+ providers that offer everything from public clouds to highly secured private clouds. Data loss in the cloud is generally lower when a cloud is used correctly but just like any other hosting model you can lose data due to human or system errors caused by human ( not automating backups, bad code, bad design, not taking advantage of HA etc ).


Don´t get me wrong. I don't have anything about the cloud model. I actually embrace it. But I think there should be millons of cloud providers and private clouds, not just 2 or 3 companies where 90% of the world puts their data. I prefer the approach VmWare is going, to develop private clouds, where small companies and clients can build their own network.


Please do more research there are many more out there today with clouds then your aware of. VMWare is the new guy in the private cloud business and even their approach is very costly or may not work at all compared to others (finicky hardware support, SAN requirements etc, VMWare licensing) make it difficult to scale out on a global footprint due to hardware availability and support.

There are other providers out there today globally that offer private clouds aka virtual private data centers (VPDC) and have been for 3 years.

Thanks Cirrostratus for your information + the original topic I must say I'm not to keen on grid/cloud solutions yet.

A. How do you know that microsoft was using "cloud" technologies, i.e. virtualization, for the storage of this data? Most likely they were using more traditional methods.

B. Even if they were using virtualization, how does the fact that HyperV is immature reflect poorly on virtualization as a whole? Do you understand that there's more than one technology that may power a "cloud"? Your argument is that HyperV (even though I think you're mistaken that virtualization was being used here in the first place) sucks, therefore we shouldn't use virtualization. That's like saying Windows Vista sucks, therefore we shouldn't use operating systems.

-- Cloud based solutions allow you to take an application (collection of vm's performing an overall service). Make it into an automated template and easily replicate it across multiple data centers. Typically at a lower cost then it would take to do the same with colo or dedicated platforms.

That's completely false. You can do the same automation quicker, simpler, and at lower cost by replicating an OS across servers. With cloud computing, you still have to do this work regardless, and then roll out your cloud layer on top in addition.


Cloud based solutions just make it easier as it will be template based and also allow you to easily snapshot your entire environment and migrate or copy it to a remote site as a backup. This of course can also then be automated to make the overall backup process easier and optionally use secure tunnels such as VPN, SSH to perform the tasks to make the overall process more secure then people today who use standard ftp based solutions.

All of this is possible without cloud.


Please do more research there are many more out there today with clouds then your aware of. VMWare is the new guy in the private cloud business and even their approach is very costly or may not work at all compared to others (finicky hardware support, SAN requirements etc, VMWare licensing) make it difficult to scale out on a global footprint due to hardware availability and support.


No matter how you look at it, cloud computing introduces more complexity. More complexity means more points of failure. Just because the complexity is hidden from the user, doesn't mean it doesn't exist.

At the end of the day, cloud computing is only of benefit to people who infrequently need a large amount of resources. Most of the time, these needs are better met by CDN.

For everyone else (the vast majority), cloud computing is unnecessary overhead. No matter how mature products and solutions become, this will never change.

By all accounts danger made a rookie mistake, somehow they did not qualify their backup solution, or test the SAN upgrade beforehand. I would normally attribute this to poor planning, simple as that. Cloud technology and terminology is mostly over-hyped marketing spin, a basic attempt to obfuscate a distributed system and the underlying technology in use. The bottom line, however, is someone wasn't doing their job, now the users suffer.

There was also this (http://www.betanews.com/article/Single-point-of-failure-blamed-for-Verizon-FiOS-DSL-outage/1254755388) issue last week, when Verizon subscribers lost internet, supposedly due to a single 'stalled' router. Most would wonder how a company like Verizon, who can afford routers and clued engineers, would be able to suffer such an outage. Just by reading some of the twitter posts (from the SVP nonetheless) you start to wonder how any of these companies provide a reasonable level of service to begin with.

Here (http://forums.verizon.com/t5/Verizon-at-Home/Northeast-Router-Failure/ba-p/87001) is the blog post from Eric Rabe (SVP) of Verizon haplessly describing the outage and supposed issue.
When routers have problems, they are designed to report that they are sick. Internet traffic is rerouted to adjacent routers automatically and sent around the trouble spot. In this case, that didn’t happen. The router went into a hung state and did not appear to the rest of the network as though it was having problems.

That meant that some user traffic from the northeast continued to flow to the stalled router, but couldn’t be processed. Presto – an outage for those users. Verizon FiOS, High Speed Internet and small business customers were affected.

The problem occurred at approximately 3:15 EDT and was detected and resolved :40 minutes later. Traffic began to flow normally and customers returned to service at that point.


After you are done scratching your head or picking yourself up off the floor from laughing too hard, ask yourself how easily avoidable both these outages were.

There is no need to enter a debate into if it was cloud or not.

There already is one here:
http://www.datacenterknowledge.com/archives/2009/10/12/the-sidekick-failure-and-cloud-culpability/

Whether some technical definition of "cloud" classifies it as such isn't really the big issue here though, is it. An average consumer doesn't know anything about SANs or virtual machines or any other technology which may or may not support a particular cloud architecture. He just knows that his data was stored "somewhere" and now it's gone, and there was nothing he could have done to prevent it except for not using it.

Once burned, twice shy.

At the end of the day, cloud computing is only of benefit to people who infrequently need a large amount of resources. Most of the time, these needs are better met by CDN.

For everyone else (the vast majority), cloud computing is unnecessary overhead. No matter how mature products and solutions become, this will never change.

Don't be silly. "Cloud" computing is a major step forward - mistakes like this have happened with non-cloud services; it's not like inadequate backup procedures are a cloud specific phenomenon, and it's disingenuous to suggest so.

CDN =! cloud hosting/computing. If you think that's all there is to the various facets of cloud computing, you should probably re-assess.

Every major website in the world runs on a "cloud" like infrastructure.

Don't be silly. "Cloud" computing is a major step forward - mistakes like this have happened with non-cloud services; it's not like inadequate backup procedures are a cloud specific phenomenon, and it's disingenuous to suggest so.

CDN =! cloud hosting/computing. If you think that's all there is to the various facets of cloud computing, you should probably re-assess.

Every major website in the world runs on a "cloud" like infrastructure.

Bump/Agree completely. What PYDOT's not seeing is how much havoc these huge websites/companies would be experiencing on non-cloud services. It'd be like claiming shared hosting is better than managed dedicated hosting. Baffling to even suggest. :eek: What is this alternative that is so much better for massively large companies to operate flawlessly on that we are comparing with?

Google's entire infrastructure is already "cloud" - from the search engine, to Gmail to Google Apps. Look up things like BigTable, MapReduce, GoogleFS etc. They're all very "cloud" technologies, and they're powering things that you already take for granted.

Google's entire infrastructure is already "cloud" - from the search engine, to Gmail to Google Apps. Look up things like BigTable, MapReduce, GoogleFS etc. They're all very "cloud" technologies, and they're powering things that you already take for granted.

Exactly, it's the next step up. That's why this whole thread is ridiculous. It'd be like pointing out one error with computers and using it as an excuse for why you should go back to the typewriter.

-- Cloud based solutions allow you to take an application (collection of vm's performing an overall service). Make it into an automated template and easily replicate it across multiple data centers. Typically at a lower cost then it would take to do the same with colo or dedicated platforms.


Bump/Agree completely. What PYDOT's not seeing is how much havoc these huge websites/companies would be experiencing on non-cloud services. It'd be like claiming shared hosting is better than managed dedicated hosting. Baffling to even suggest. :eek: What is this alternative that is so much better for massively large companies to operate flawlessly on that we are comparing with?

By your cloud definition im running on cloud for at least 5 years.

If you define Xen, VmWare or any virtual solution in a cluster environment, then cloud exists for at least 10 years.

I defined Google, Amazon etc, as cloud because they are truly different then the approach you just called cloud. Snapshots? Backups? That is not cloud as far as I concern, then we all run on clouds already. At least the ones that use virtualization then can start to market their services as Cloud by your definition.

On a real cloud, you would never need templates and snapshots like you described. Thats just virtualization and nothing more. Cloud uses virtualization but its a more complicated setup then just running some virtual boxes.

On a cloud system your data is replicated or mirrored on several servers at the same time, just like a RAID setup does with disks, servers are disks in a cloud, and your data is elastic, its on several servers at the same time, where several physical boxes can fail and your system would be still be running without snapshots, or backups at all. The minute you need to take a snapshot and move it to another device you have a point of failure. Even SAN devices fail. So by my definition cloud is where your data is already mirrored on several parts and that just leaves a few providers, Google, Amazon and some more.

Whats next. Setting a OpenVZ server and call it cloud?

I'm not against cloud at all. I said that before here, this topic is just about discussion and debating, its not about if cloud is bad or not. Cloud is nothing more then a new marketing term. The technology existed years before everybody started to talk about it, before this cloud bubble. You said every major website runs on cloud, and you are right, and thats just proofs the point. Where did they run those websites before this cloud era? They did on the same systems they do now, even 10 years ago, before all this cloud explosions. Google runs for more then 10 years and so do allot of others, which used a grid/cluster system before. What happened? Wasn't cloud invented yet? Of course it was, it just had another name.

By your cloud definition im running on cloud for at least 5 years.

If you define Xen, VmWare or any virtual solution in a cluster environment, then cloud exists for at least 10 years.

I defined Google, Amazon etc, as cloud because they are truly different then the approach you just called cloud. Snapshots? Backups? That is not cloud as far as I concern, then we all run on clouds already. At least the ones that use virtualization then can start to market their services as Cloud by your definition.

On a real cloud, you would never need templates and snapshots like you described. Thats just virtualization and nothing more. Cloud uses virtualization but its a more complicated setup then just running some virtual boxes.

On a cloud system your data is replicated or mirrored on several servers at the same time, just like a RAID setup does with disks, servers are disks in a cloud, and your data is elastic, its on several servers at the same time, where several physical boxes can fail and your system would be still be running without snapshots, or backups at all. The minute you need to take a snapshot and move it to another device you have a point of failure. Even SAN devices fail. So by my definition cloud is where your data is already mirrored on several parts and that just leaves a few providers, Google, Amazon and some more.

Whats next. Setting a OpenVZ server and call it cloud?

I'm not against cloud at all. I said that before here, this topic is just about discussion and debating, its not about if cloud is bad or not. Cloud is nothing more then a new marketing term. The technology existed years before everybody started to talk about it, before this cloud bubble. You said every major website runs on cloud, and you are right, and thats just proofs the point. Where did they run those websites before this cloud era? They did on the same systems they do now, even 10 years ago, before all this cloud explosions. Google runs for more then 10 years and so do allot of others, which used a grid/cluster system before. What happened? Wasn't cloud invented yet? Of course it was, it just had another name.

Sorry for the late response, been AFK for a week. I don't disagree with you that it's kind of lame that everyone is calling it "cloud" computing these days. But that'd be like arguing against all these web hosting companies calling themselves "green", when all they are doing is using a new marketing buzz word, giving maybe 1% of revenue for a revenue boost that far exceeds it (because some green consulting firm convinced them of the "ROI" of going "green" by purchasing REC's) . That's life.

But your whole thread was designed in a way to say "cloud computing is dangerous". Like I said before, what is this better alternative if cloud is so bad? I'm actually curious, is there any "alternative" that doesn't label itself cloud that would accomplish what these companies want? No, of course not. At this point the term has become so ambiguous it's almost used simply as the "next step" beyond dedicated hosting, or using their own infrastructure for many, instead of utility computing which is what most of them are. Many a person argues over whether it is simply utility/grid computing repackaged or whatnot.

Regardless, to declare "cloud computing can be dangerous" presumes there is some sort of alternative that is *not* dangerous. And since as we discussed cloud computing is a broad concept, that is a rather irrational claim. In all seriousness though, what is a "not dangerous" alternative?

In my opinion, everything can be dangerous if you are not properly planning for the worse. I get asked every day, "If this cloud is HA, then I don't have to keep backups, right?" WRONG. There will never be a situation when that is true, no matter how promising the platform may be. The reason we have versions of software is because there are always room for improvement and there are always bugs that have not been discovered. Last thing you want is to discover a bug that has wiped out all your data. Even if you have the provider keeping backups for you, that is not enough. Keep your own backups. You never know what can happen.

If you are looking to move onto a cloud solution, always do your research and make sure you are choosing the right platform that is managed by the right people. The cloud platform is only as good as the quality behind it - network, support and fail-safe.

This was bad althought its back up and all, I lost all my photos and emails and important information on my phone. Yes, I have a sidekick LX. And it sucks

I agree that a true Cloud setup would involve replication across more than one server. I don't understand what the danger warnings are all about. You can lose data in a traditional Server/Hosting structure just as easily in the Cloud grid. Data Backup and Disaster Recovery have always been a part of the process, Clouds are not immune. So what.

I think there is a very dangerous precedent set with the term "Cloud" in that there isn't a consistent definition, so arguments about what it means spring up all over the place - and worse, assumptions are made about what a supplier means when they offer a cloud service.

The cloud platform is only as good as the quality behind it - network, support and fail-safe.

I concur - besides, these systems are still run (and the software written) by people, and it's possible for even the best of them to make mistakes.

...As far as I concern cloud exists for years, it was called grid before, they just added the pay per consume model.

Can somebody verify or add to this guy's understanding of Cloud Computing?

I've always had a problem understanding the term.

If CC is simply just Grid Computing with a pay-per-consume or subscription based model, that really just simplifies things for me...

SaaS is easier to understand, and since SasS, Cloud Computing and Green technology seem to be the latest things, I'm still pondering on cloud computing, as I'd love to get into all these new trends.

Really would appreicate anybody's input...

This is what happens when you live without internet connectivity at home for 2 years, so lost touch...

-----

To me, prime examples of Cloud computing (and SaaS) are Facebook and Twitter.
I can see advantages to the grid computing and utility computing concepts of these examples, but of other businesses? What other examples of cloud computing in particular are there?

I know that Kayako's products are great examples of SaaS... But it's stillt he Cloud computing thing, after reading Pydot's understanding, and from my own understanding (however little) I had just assumed "cloud computing" was just a buzz word for marketing for a grid/utility concept (and pay per consume).

Sigh, edit button expired,
wanted to continue with:

"I know that Kayako's products are great examples of SaaS... But it's stillt he Cloud computing thing, after reading Pydot's understanding, and from my own understanding (however little) I had just assumed "cloud computing" was just a buzz word for marketing for a grid/utility hybrid concept (and pay per consume).

After all, "Grid Utility Hybird Servers" would be a very long winded way of saying things."

Apologies for repeated messages...

But I just got a peelback ad on screen for Carinet...
Out of curiousity I checked it out and because it was on topic I checked out their "Grid & Cloud" page.

First paragraph on page:
"Grid Utility Hosting On The Cloud
There, now that we have all the “buzzwords” out of the way, we can get to the business of it all. Grid/Cloud/Utility computing/hosting is the same thing. CariNet stands by the following definition: “a type of computing where dynamic resources are provided as a service over the Internet."

Link: Grid Hosting VPD (http://www.cari.net/grid-hosting-vpd.html)

Well, if anybody wants to argue that, by all means, but I kind of shot myself reading that.

Again, apologies for "spamming" I'll shut up now...

There is no definition, that's half the problem. Different companies and people have their own interpretations of what features are required as a cloud hosting provider.

http://bobdoeshosting.blogspot.com/2009/11/what-does-cloud-mean.html

I did not know that could happen from Cloud computing,
So bad that many have lost their data.

This is the most useless post/article I've ever read. People should take responsibility for their data and make off-site redundant backups if it is that important to them.

Everyone on here is so concerned about cloud failure and how it's going to cost them their business. My question to you is why in the world would you put your ENTIRE business on a cloud and not have some type of contingency plan? If you don't then you deserve to suffer a catastrophic data loss. T-Mobile is dumb for not having thought of this ahead of time. Don't blame the technology. Blame the dumb admins who made the decision to put all their eggs in one basket with no backup of their own.