In realm of big IT where I have about 13PB of backup data on deduplicated disk we didn't even look at these products. Data Domain's DD690 and 880 are overall excellent but can't compress Oracle data if their lives depended on it. At least not the ways that the DBAs like to back up Oracle. IBM's Diligent product is a fantastic piece of technology for both Open and Mainframe systems, but is VTL based and does not come cheap.
Optimized replication between sites is one of the best parts of dedupe, even over storage. If something can actually get 10x compression then that 1GbE link I have between locations functionally acts more like a 10GbE for no more cost. A huge boost on the WAN for DR.
One of the things that wasn't talked much about in the press conference was that the software heavily utilizes the GPU over the CPU when compatible hardware exists. I meant to bring it up somehow, but I was happy to be done and off camera after an hour. Media events, while interesting, require a lot of sitting still, being quiet, and not sneezing.
Yes, the technology for doing distributed computing is now over ten years old and I was a very early adopter. So as some people pointed out that's not new 'news' anymore per say. What is computationally newer is that the projects now don't just expand at Moore's law's rate anymore and as GPUs get better it will increase much faster for the next few years until leveling off at some new growth rate. Yes I know other things have been found, but finding a pulsar was really cool. Speaking with the scientists and science media all over the world and seeing the full international scope of this project over the last few weeks was also fascinating.
With all of the talk about SPARC (I use and love the new 8 core SPARCs) there hasn't been an talk of Sun's StorageTek disk and tape drives. The 4GbFC, 1TB T10000K drive is critical for enterprise storage. Legacy support for 9940s and 9840s are also critical for big iron. I don't like the idea of having to move to IBM 1130s. Sun SATA JBOD arrays are excellent tier 3 storage, they can move a lot of IO for not a lot of cash. Put a Sun 5220 server (or Mx000) with multiple 4Gb FC HBAs in front of SATA using ZFS and you can build a massive data warehouse or backup system.
Why would someone just buy drives like that? If there is that large of a need for space, go buy a good array from Sun, NetApp, Hitachi, IBM, EMC, or a smaller vendor. Let them do the testing, warranty work, and integrity and let you work on your business. What do you do if you need more space than a DAS RAID can fit?
With iSCSI, FCIP, FOE, 10GbE, and FC being much cheaper/free than years ago why use DAS disk? Slice those big drives into RAID 6 and then into whatever sized LUNs you need. 5GB for a boot? Well you can fit 200 onto a RAID group. Need a 5TB Lun with good speed? Grab 1TB from several different RAID groups and have the array controller stripe for you. 8Gbs FC is out now, 2Gbs FC is really cheap with used equipment. iSCSI is pretty much free. You can even virtualize whole arrays so if you give someone 1TB and they only use 5Gb, only 5Gb is actually in use on physical disk. If you buy a deduplicating array it'll even eliminate redundant information for you depending on how you set it up.
If you value stability, flexibility, and redundancy and don't like employees wasting lots of time on hard drive testing and fixing, go buy the right tool for the job. Large arrays are also usually 'greener' since they can use drives much more efficiently if set up correctly. One 15K RPM 300GB FC RAID set at full IO load can beat hundreds of cheap SATA drives at low utilization for IO. If space is there problem then there are ways to work that too.
This must have been what my Sun sales guy was talking about a month ago at lunch when he had a product that he described that was very similar to this, but he couldn't even give me the full details. I wanted to buy about a PB of then then and there, but we'll see what my other vendors come back with as counters. The new HDS arrays are changing to SAS from FC and have excellent virutalization on them. I've been pushing for 10K SATA2 drives to really shake up enterprise level storage. I'd love to have these mixed into the ~6PB of disk I help design and plan for.
IOPS and repose time are the keys in storage, not the actual cost or specific technology. Just like the space shuttle main computers, it doesn't matter if they cost a lot or are old. They have to be right 100% of the time and do what they need to do in the time allowed.
Holy cow, it's nice to know that I'm not the only person who has a Tandem system in my datacenter. Hum, that makes me wonder if we are in the same place.
There is actually a good chance this is oracle data or MS SQL. I work for a F100 company doing backup and storage, and most of our data is not mainframe based. However, the way modern tapes interleave data, compress it, and break it up across tapes, it does make it hard to read back if you don't know which of the five or six enterprise backup software companies made the backup or the order of the data. For all of you disk backup fanboys out there who don't work in the trans-PB world, it's hard and very, very expensive to keep the 1.5PB of backup data I have on disk or replicate it somewhere. 500GB 120MB/s tapes are still the cheapest storage for most backup data. For all you 'green' people out there, tapes are way more power effective than spinning thousands of 15K FC or 7.2K SATA drives.
As for why everyone doesn't use encryption... imagine the cost of converting several PB of different technologies and ages to use encryption and then be able to read it back in the seven, 30, or infinite years the government or lawyers require us to keep some records. It's a many-multi-million dollar problem with no good solution.
Here in Wells' IT, we'd like to encourge you to take out the new loan so we can afford to buy a disk array of these things to keep track of your loans.
Drives are cheap, it's the SAN hookups that are expensive. Our 80TB Sun array (Tier II Storage) is full of the same Hitachi disks that I have in my desktop. However, Sun provides the advanced controlers and software to make sure we have zero downtime.
Our cisco guy won't even sell it to us, he said he wants to keep us as a customer. Without naming names, we're a Fortune 100 company and are the target market for this.
I'm able to write a data stream to my newer tape drives at over 120MB/s. Thanks to hardware compression I can hit my 30MB/s drives at 80-90MB/s depending on the data. Tape drives are good at what they do, and don't call them slow. They are just serial storage istead of random access. My tapes are faster than almost anything else on my SAN and LAN, just a different method of cheaply storing over a PB of data.
If IBM mainframes could talk to my 40GB ipod, we could get rid of our shelves and shelves of 210MB data tapes. My company hasn't figured out the pain it would take to attach new T10K drives in emulation mode to the mainframe would pay for itself very quickly since we'd get to re-purpose an operator or two.
If you are already using NetBackup or BackupExec on your servers, just extend what you are doing with the Desktop/Laptop Agent. I use it at home with BExec 10 and it works well for my wife's laptop, it syncs when she is at home. I work for a very large bank, and we use it a bit with NetBackup. For the most part we forbid any data to leave the datacenter, but that is up to each company to determine their risk/profit tolerance.
A standared dual CPU dual core HP server with Windows can keep a 4Gb FC pretty full if set up correctly. I work for a large bank, and we have many a Solaris box that can keep 4 or even 8 2Gb FC cards full into our FC and SATA disk arrays. Not to trivialize the extreme coolness of what they are doing at all, but a PB of data with a few PB of I/O in a day isn't what it used to be. I'm just glad to see they don't use Polyserve, it is worthless for clustering and has caused more downtime at work than it has ever prevented. If they really have that much data they should use 10Gb FC or Infiband. Even our stodgy old bank is implementing our first infiband system so we can move IO at 12Gb instead of the slow 4Gb links.
My favorite PC is still running DOS 3.2. It worked great, it never crashes, I have no network problems since there isn't one. My 286 provides enough horsepower to run WP5.1, Norton, and Gorillas.
Where are these rich corn farmers you speak of? I'll ask my father, uncles, and grandfather if they are rich. Or any of the other corn farmers I know here in Iowa if they know any. I guess they've all been holding back on me all of these years...
They cost about $30,000/ea. Our solution is the 500GB/tape raw, STK T10000A drive, they write at 120MB/s raw on our 2Gb FC and get about 1.67:1 compression on our 400 servers. The T10000B will use 1TB raw tapes and have 4Gb fiber interfaces.
Quite a bit of enterprise software and hardware uses this model alrady. With Hitachi and Veritas we need a license key for every single little bit of the software, and often times another one to enable it per 1 or 10 TB of storage. Hitachi probably provides the hardware at a loss to them (still expensive to us), but they make their money on their software.
One tape drive, why stop there, for $200 you can get a 10 tape robot on ebay for DLT IV tapes. I have 10, 35GB tapes in mine ("Bender") right now. For my small server and workstations. I have problems with a stuck tapes once in a while, and my wife thinks I was nuts to by the thing, but it works really well. For taking data off site, tapes are still the best thing out there.
I just wish I had a fiber 3590K, 9940B, or T10000 drives like I have at work for home.
I also did my master's thesis on a concept like this. I was doing this splitting and syncing using virtual disks. They communicated on AES encrypted TCP packets. I idea was it would be a system simple enough for any to set up as a bunch of friends or families.
Slashdot Mirror
In realm of big IT where I have about 13PB of backup data on deduplicated disk we didn't even look at these products. Data Domain's DD690 and 880 are overall excellent but can't compress Oracle data if their lives depended on it. At least not the ways that the DBAs like to back up Oracle. IBM's Diligent product is a fantastic piece of technology for both Open and Mainframe systems, but is VTL based and does not come cheap.
Optimized replication between sites is one of the best parts of dedupe, even over storage. If something can actually get 10x compression then that 1GbE link I have between locations functionally acts more like a 10GbE for no more cost. A huge boost on the WAN for DR.
One of the things that wasn't talked much about in the press conference was that the software heavily utilizes the GPU over the CPU when compatible hardware exists. I meant to bring it up somehow, but I was happy to be done and off camera after an hour. Media events, while interesting, require a lot of sitting still, being quiet, and not sneezing.
Yes, the technology for doing distributed computing is now over ten years old and I was a very early adopter. So as some people pointed out that's not new 'news' anymore per say. What is computationally newer is that the projects now don't just expand at Moore's law's rate anymore and as GPUs get better it will increase much faster for the next few years until leveling off at some new growth rate. Yes I know other things have been found, but finding a pulsar was really cool. Speaking with the scientists and science media all over the world and seeing the full international scope of this project over the last few weeks was also fascinating.
1GbFC? How well can this stand up to modern 8GbFC or 10GbE iSCSI?
With all of the talk about SPARC (I use and love the new 8 core SPARCs) there hasn't been an talk of Sun's StorageTek disk and tape drives. The 4GbFC, 1TB T10000K drive is critical for enterprise storage. Legacy support for 9940s and 9840s are also critical for big iron. I don't like the idea of having to move to IBM 1130s. Sun SATA JBOD arrays are excellent tier 3 storage, they can move a lot of IO for not a lot of cash. Put a Sun 5220 server (or Mx000) with multiple 4Gb FC HBAs in front of SATA using ZFS and you can build a massive data warehouse or backup system.
Why would someone just buy drives like that? If there is that large of a need for space, go buy a good array from Sun, NetApp, Hitachi, IBM, EMC, or a smaller vendor. Let them do the testing, warranty work, and integrity and let you work on your business. What do you do if you need more space than a DAS RAID can fit?
With iSCSI, FCIP, FOE, 10GbE, and FC being much cheaper/free than years ago why use DAS disk? Slice those big drives into RAID 6 and then into whatever sized LUNs you need. 5GB for a boot? Well you can fit 200 onto a RAID group. Need a 5TB Lun with good speed? Grab 1TB from several different RAID groups and have the array controller stripe for you. 8Gbs FC is out now, 2Gbs FC is really cheap with used equipment. iSCSI is pretty much free. You can even virtualize whole arrays so if you give someone 1TB and they only use 5Gb, only 5Gb is actually in use on physical disk. If you buy a deduplicating array it'll even eliminate redundant information for you depending on how you set it up.
If you value stability, flexibility, and redundancy and don't like employees wasting lots of time on hard drive testing and fixing, go buy the right tool for the job. Large arrays are also usually 'greener' since they can use drives much more efficiently if set up correctly. One 15K RPM 300GB FC RAID set at full IO load can beat hundreds of cheap SATA drives at low utilization for IO. If space is there problem then there are ways to work that too.
This must have been what my Sun sales guy was talking about a month ago at lunch when he had a product that he described that was very similar to this, but he couldn't even give me the full details. I wanted to buy about a PB of then then and there, but we'll see what my other vendors come back with as counters. The new HDS arrays are changing to SAS from FC and have excellent virutalization on them. I've been pushing for 10K SATA2 drives to really shake up enterprise level storage. I'd love to have these mixed into the ~6PB of disk I help design and plan for.
IOPS and repose time are the keys in storage, not the actual cost or specific technology. Just like the space shuttle main computers, it doesn't matter if they cost a lot or are old. They have to be right 100% of the time and do what they need to do in the time allowed.
And if you really want it to work, run Solaris or zOS.
Holy cow, it's nice to know that I'm not the only person who has a Tandem system in my datacenter. Hum, that makes me wonder if we are in the same place.
There is actually a good chance this is oracle data or MS SQL. I work for a F100 company doing backup and storage, and most of our data is not mainframe based. However, the way modern tapes interleave data, compress it, and break it up across tapes, it does make it hard to read back if you don't know which of the five or six enterprise backup software companies made the backup or the order of the data. For all of you disk backup fanboys out there who don't work in the trans-PB world, it's hard and very, very expensive to keep the 1.5PB of backup data I have on disk or replicate it somewhere. 500GB 120MB/s tapes are still the cheapest storage for most backup data. For all you 'green' people out there, tapes are way more power effective than spinning thousands of 15K FC or 7.2K SATA drives.
As for why everyone doesn't use encryption... imagine the cost of converting several PB of different technologies and ages to use encryption and then be able to read it back in the seven, 30, or infinite years the government or lawyers require us to keep some records. It's a many-multi-million dollar problem with no good solution.
Here in Wells' IT, we'd like to encourge you to take out the new loan so we can afford to buy a disk array of these things to keep track of your loans.
Drives are cheap, it's the SAN hookups that are expensive. Our 80TB Sun array (Tier II Storage) is full of the same Hitachi disks that I have in my desktop. However, Sun provides the advanced controlers and software to make sure we have zero downtime.
SCSI? What sort of backwards datacenter are you in? We (and our peer companies) use 95%+ FC or SATA2. SCSI drives were so 1990s....
Our cisco guy won't even sell it to us, he said he wants to keep us as a customer. Without naming names, we're a Fortune 100 company and are the target market for this.
I'm able to write a data stream to my newer tape drives at over 120MB/s. Thanks to hardware compression I can hit my 30MB/s drives at 80-90MB/s depending on the data. Tape drives are good at what they do, and don't call them slow. They are just serial storage istead of random access. My tapes are faster than almost anything else on my SAN and LAN, just a different method of cheaply storing over a PB of data.
If IBM mainframes could talk to my 40GB ipod, we could get rid of our shelves and shelves of 210MB data tapes. My company hasn't figured out the pain it would take to attach new T10K drives in emulation mode to the mainframe would pay for itself very quickly since we'd get to re-purpose an operator or two.
If you are already using NetBackup or BackupExec on your servers, just extend what you are doing with the Desktop/Laptop Agent. I use it at home with BExec 10 and it works well for my wife's laptop, it syncs when she is at home. I work for a very large bank, and we use it a bit with NetBackup. For the most part we forbid any data to leave the datacenter, but that is up to each company to determine their risk/profit tolerance.
A standared dual CPU dual core HP server with Windows can keep a 4Gb FC pretty full if set up correctly. I work for a large bank, and we have many a Solaris box that can keep 4 or even 8 2Gb FC cards full into our FC and SATA disk arrays. Not to trivialize the extreme coolness of what they are doing at all, but a PB of data with a few PB of I/O in a day isn't what it used to be. I'm just glad to see they don't use Polyserve, it is worthless for clustering and has caused more downtime at work than it has ever prevented. If they really have that much data they should use 10Gb FC or Infiband. Even our stodgy old bank is implementing our first infiband system so we can move IO at 12Gb instead of the slow 4Gb links.
My favorite PC is still running DOS 3.2. It worked great, it never crashes, I have no network problems since there isn't one. My 286 provides enough horsepower to run WP5.1, Norton, and Gorillas.
Where are these rich corn farmers you speak of? I'll ask my father, uncles, and grandfather if they are rich. Or any of the other corn farmers I know here in Iowa if they know any. I guess they've all been holding back on me all of these years...
They cost about $30,000/ea. Our solution is the 500GB/tape raw, STK T10000A drive, they write at 120MB/s raw on our 2Gb FC and get about 1.67:1 compression on our 400 servers. The T10000B will use 1TB raw tapes and have 4Gb fiber interfaces.
Quite a bit of enterprise software and hardware uses this model alrady. With Hitachi and Veritas we need a license key for every single little bit of the software, and often times another one to enable it per 1 or 10 TB of storage. Hitachi probably provides the hardware at a loss to them (still expensive to us), but they make their money on their software.
One tape drive, why stop there, for $200 you can get a 10 tape robot on ebay for DLT IV tapes. I have 10, 35GB tapes in mine ("Bender") right now. For my small server and workstations. I have problems with a stuck tapes once in a while, and my wife thinks I was nuts to by the thing, but it works really well. For taking data off site, tapes are still the best thing out there.
I just wish I had a fiber 3590K, 9940B, or T10000 drives like I have at work for home.
Because then we'd be able to see Amy's obscene tatoo.
If you give enough people the ability to gain economic freedom, then a few years down the road they might demand political freedom as well.
Small world,
I also did my master's thesis on a concept like this. I was doing this splitting and syncing using virtual disks. They communicated on AES encrypted TCP packets. I idea was it would be a system simple enough for any to set up as a bunch of friends or families.
I've been too busy to work too much on it since graduating, but I opened all of the code and it's on sourceforge.
http://sourceforge.net/projects/netraid/