Tape Library

Tape Library

Cyber resilient, energy-efficient tape storage with airgap and long-term retention

 View Only

Mainframe Disk and Tape Announcements

By Tony Pearson posted Tue October 21, 2008 12:59 AM

  

Originally posted by: TonyPearson


Well, it's Tuesday again, and that means more announcements from IBM!

In conjunction with IBM's new [System z10 Business Class (BC)] mainframe designed for Small and Medium-sized Businesses (SMB), IBM also announced related storage productenhancements.

DS6000 series

Yes, it's alive! Contrary to the FUD you might have read from our competitors, IBM continues to sell thousands and thousands of IBM System Storage DS6800 disk systems, and now enhances them with the optionfor 450GB 15K RPM drives. What is nice about these 450GB drives is that they are as fast or faster* than 300GBdrives, so the typical trade-off between performance and capacity do not apply.

(* I compared Seagate 15.6K (450GB) with 15.5K (300GB) models.
450GB drives300GB drives
Avg Seek time (Read)3.4ms3.5 ms
Avg Seek time (Write)3.9ms4.0 ms
Full Seek time (Read)6.43ms6.7 ms
Full Seek time (Write)7.12ms7.4 ms
Sustained Bandwidth112-171 MB/sec73-125 MB/sec

This may or may not result in application performance improvements, depending on workload pattern. Your mileage may vary.)

Our clients report back that these are incredibly stable systems that they don't have toworry about. This enhancement applies to both the [511/EX1 models] and [522/EX2 models].

DS8000 series

Understanding that clients want complete solutions from single vendors, IBM offers synergy between System z and the IBM System Storage DS8000 disk systems. The latest R4.1 microcode upgrade offers two key features onthe various models [2107,



2421, 2422, 2423, and 2424].

  • zHPF - High Performance FICON for System z. IBM was able to increase the throughput on 4 Gbps links. For OLTP workloads randomly accessing 4KB blocks, IBM internal tests showed zHPF doubled performance from 13,000 IOPSto 26,000 IOPS per channel. For sequential workloads, such as batch processing, zHPF increased performance 50 percent, from 350 MB/sec to 525 MB/sec.

  • In February, IBM previewed[IncrementalResync] for z/OS Metro Global Mirror. However, some concepts are better explained with pictures.


    One way to set up a 3-site disaster recovery protection is to have your production synchronously mirrored to a second site nearby, and at the same time asynchronously mirrored to a remote location. On the System z, you can have site "A" using synchronous IBM System Storage Metro Mirror over to nearby site "B", and also have site "A" sending data over to site "C" asynchronously using z/OS Global Mirror. This is called "z/OS Metro Global Mirror".

    In the past, if the disk system in site A failed, you would switch over to site B, which would have to resend send all the data again to site C to be resynchronized. This is because site B was not tracking what the System Data Mover (SDM) reader had or had not yet processed.

    With DS8000 4.1, the "incremental resync" function that, along with using IBM HyperSwap, requires site B to only send and resync the data that was in-flight when the outage occurred. When you compare the difference in sending this limited amount of in-flight data with the traditional complete volume of data, you can see how "Incremental Resync" can resynchronize the data 95% faster, and also greatly decrease your bandwidth requirements. This reduces the risk in case a subsequent outage occurs.



TS7700 series

Introduced originally in 1997 as the IBM Virtual Tape Server (VTS), the [IBMSystem Storage TS7700] series supports Grid capabilityto replicate tape image data across locations. Here's a quick recap of today's announcement:

  • Existing TS7740 can be upgraded up to 9TB of disk cache. New models can have up to 13TB of disk cache.
  • A new "tape-less" TS7720 that has up to 70TB of disk cache.
  • Integrate Library Management support. I discussed[IntegratedRemovable Media Manager (IRMM)] before, and this is basically IRMM inside. For those with TS3500 tape libraries,this support eliminates the need for a separate IBM 3953 L05 Library Manager.
  • TS1130 back-end tape drive support. These are the fastest 1TB drives in the industry, with support of built-in encryption, and now can be used asthe physical tape back-end for the virtual tape TS7740 repository.

While our competitors might be boarding up their windows in preparation for the economic downturn in the USAeconomy, IBM remains generating solid results. San Jose Mercury News has an article that discusses this titled[IBM's 3Q profit strong on global sales].There has never been a better time to buy from, or invest in, IBM!

technorati tags: , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,

2 comments
6 views

Permalink

Comments

Tue October 21, 2008 05:35 PM

Chuck,I compared Seagate models based on their DDM speeds, I had not looked at the HGST, but glad to hear that their 450GB are as fast as their 300GB also. The magic caching algorithms and design features in the DS6800 and DS8000 apply to both types of drive. Depending on the workload, these higher capacity drives could therefore help deliver the same or more IOPS and MB/sec throughput as the 300GB drives. I have updated to show the numbers I used for the comparison to clarify.
I have also clarified the Incremental Resync paragraph.-- Tony

Tue October 21, 2008 04:13 PM

Tony -
How can you claim that the 450 GB drives are as fast or faster than the 300 GB drives and then conclude that capacity/performance trade-offs don't matter?? I checked HGST drive specs, the 300 and 450s have identical buffer sizes, latency and seek times. I think each drive can sustain just under 180 IOPS per drive (clear this up for me if I have missed something) and that the capacity/performance trade-off is alive and well. Is there some magic that happens inside a DS6800 or DS8000 that allow a 450 GB drive to handle more IOPS than a 300 GB drive?