Thursday, April 7, 2011

IBM Data Power on zBX

I covered this already in our newsletter, but I still want to repeat it here as this whole zEnterprise story really seems to mark a turning point in our mainframe history.

I'll start off with a video, just watch it before you continue reading.

OK, now you may ask yourself : "how did I get here, am I right am I wrong". Oops, no, that's Talking Heads. Second attempt : now you may ask yourself : "what does this have to do with me ?" Actually, a lot ! The IBM Websphere DataPower Integration Appliance XI50 for zEnterprise (XI50z) is - next to the Smart Analytics Optimizer - announced as the following optimizer on zBX. And it fits perfectly in the new zEnterprise strategy of IBM.

Let me repeat some of my own words when I discussed the announcement of the z196 EC : "Starting point for this is that we live in a heterogeneous world where a one size fits all approach for all workloads does no longer stand a chance. Each platform has its specific strengths. What IBM wants to do now is bring those platforms together into one system, the IBM zEnterprise System. The goal is to use the mainframe as a central system to manage the heterogeneous platforms which are connected to it." I think the XI50z is quite a good example of this strategy as it bring's an existing appliance to the mainframe. And there are good reasons for doing so. Let me tell you some more about it.

Some history
The DataPower appliances (there are several) were added to the IBM portfolio in 2005 when IBM took over DataPower. De XI50z is actually the System z version op the existing XI50B appliance, you also see in the video above. These appliances are hardware platforms built with a specific goal : functionalities that are originally executed at server level are consolidated into appliances. This does not only consolidate these functionalities, it should also boost your performance.

Click on image for larger version in new window

How does it work ?
It's clear that this follows the same philosophy as what I described above or it's at least a predecessor of it. Summarized the XI50z is specifically used to accelerate XML processing, converting Web services protocols and also taking up some security aspects.

Ok, it's clear, you can best describe this as an Enterprise Server Bus (ESB). Let's give you some functionalities in more detail :
  • Protocol Bridging : Integrate disparate transport protocols with extreme ease like : HTTP(s), WebSphere MQ, WebSphere JMS, Tibco EMS, Database (DB2, Oracle, Sybase, SQL Server)
  • "Parsing of messages to allow specific routing based on Quality of Service requirements by user or transaction type.
  • The ability to take synchronous or asynchronous action based on the message, such as additional database calls to add to the message, or to modify the message content.
  • Conversion of XML to more efficient protocols than the SOAP or HTTP format, such as WebSphere MQ."
As I also indicated, IBM states the use of these appliances should give you a performance gain. For the XI50 this should be around 23%.

What about the mainframe
Here are some functionalities the XI50z (or for that matter of course also the XI50B) can handle in a mainframe context :
  • Connect to existing applications over WebSphere MQ
  • Transform XML to/from COBOL Copybook for legacy needs
  • Natively communicate with IMS Connect
  • Dynamic crypto material retrieval & caching, or offload crypto ops to z
  • Service enable CICS using WebSphere MQ
  • Virtualize CICS Web Services
As always, a picture tells more than a thousand words, so here are a couple of examples for CICS and DB2
Click on image for larger version in new window

Click on image for larger version in new window

Integration of the DataPower appliance on zBX
For the technical implementation I'm going to refer to the announcement, but I would like to point out some of the advantages for putting this appliance on mainframe (zBX) instead of running it as a standalone appliance.

Blade Hardware Management
  • Monitoring of hardware for health, degraded operation
  • Call-home for current/expected problems, automatic dispatch of CSR
  • Consolidation/Integration of DP HW problem reporting with other problems reported in zBX
  • Energy Monitoring and Management of DP Blades
DataPower Firmware Load and Update
  • Consistent change mgmt with other zGryphon firmware mgmt
  • Enforced restriction of firmware updates to SE userid
  • Enhanced new firmware level testing in zBX
HMC Console Integration
  • Person monitoring the z environment from an overall hardware operational perspective will see DP blades included in the picture, with associated status from a single console
  • Group GUI operations for functions supported on HMC (e.g. power up/quiesce/upgrade firmware)
  • Time synchronization with system z time via HMC/SE time server
DataPower Failure Recovery and Restart
  • HMC/SE will detect and report on appliance failures and can be used to re-cycle appliance if DataPower built-in restart fails
  • Periodic Backup/restore of full blade configuration (automatic on changes to config); Backup to HMC media
  • Virtual Network Provisioning
  • Provides enforced isolation of network traffic via VLAN support
  • 10Gb end-to-end network infrastructure
  • Built-in network redundancy
  • IEDN provides protected network, possibly obviating customer-perceived need for encryption of last-mile flows between DP and target back-end server
Monitoring and Reporting
  • Monitoring of DataPower health via HMC
  • Consolidated platform error logging across whole environment
I repeat my question : "what does this have to do with me ?". Well, in the coming years we will see more and more of these 'strange' technologies coming to the mainframe. All technologies that wé are not familiar with. Still, we will have to invest time in them, learn to know them because otherwise we will never be able to convince people that these technologies can run on, yes, belong on the mainframe. This is a new challenge we are facing. Let this XI50z be a first testcase for you.

I'll end up with another video, specifically on the XI50z and how you can implement it on mainframe.

1 comment:

Anonymous said...

monitoring and healthcheck of zBX and its blade servers should be available through a daemon on z/OS, but not on HMC. No standard operating and escalation procedures and be linked into this