I already have a Parallel Sysplex spread across multiple sites and already use synchronous disk copy. Why do I need anything else?
There are several reasons:
1) Synchronous disk remote copy by itself can not provide data consistency. There are many examples of how individual secondary disk can drop out of duplex mode. Without consisency, the data is useless. GDPS is designed to guarantee data consistency for the System z data, together with distributed data.
2) Recovering from a real disaster involves a lot of steps including managing activation profiles, CBU, CF structures, remote copy configuration, IPLing, Couple Data Sets, etc. GDPS is designed to automate the entire process, requiring only a single confirmation by an operator. This effectively removes key personnel as a "Single Point of Failure" in your disaster recovery plans.
What does GDPS stand for?
GDPS stands for "Geographically Dispersed Parallel Sysplex," but GDPS has grown by adding additional function such as unlimited distance for disaster recover. GDPS can manage the high availability sysplex within a metro area and the recovery ability of that same sysplex at a location at unlimited distance. Because of these expanded capabilities, it is just called by the name "GDPS."
Why "parallel sysplex" when we're discussing long distance, asynchronous disaster recovery?
For "metropolitan" distances, a z/OS Parallel Sysplex with data sharing provides significant availability benefits. GDPS can manage any sysplex environment and does not require a parallel sysplex. IBM chose to use the Parallel sysplex in this product name because the GDPS offering was and still is designed to deliver the highest possible availability for the IT environment.
Since today GDPS can extend well beyond the limits of a z/OS Parallel Sysplex, we have renamed to product to just "GDPS."
Are Coupling Facilities mandatory?
Although a "base" syplex (without a CF) is required for GDPS/PPRC and GDPS/PPRC HyperSwap Manager. Data sharing, enabled with a Parallel Sysplex, provides significant availability benefits. Resource Sharing enables system management and performance benefits with cost savings. Because of this, we highly recommend a Parallel Sysplex where appropriate.
I have to buy Tivoli System Automation. Are there any other options?
The Tivoli SA for z/OS is the only automation product that exploits the capability of the Parallel Sysplex environment. All other automation products are only sysplex tolerant, being only single image automation products. The major benefit of GDPS is to fully automate the day to day infrastructure management and the disaster recovery processes. The GDPS automation is based upon Tivoli System Automation for z/OS because of its functionality and ability to manage the whole sysplex from a central point. This very powerful product has pre-written automation routines to cover a wide range of activities. GDPS does provide the capability to coexist with other automation products. This means that you can implement the basic functions that GDPS must have in place and then turn the automation control over to the preexisting automation product that a client has in place. GDPS/PPRC HyperSwap Manager comes with an option to run with a special reduced function (and price) version of SA and for NetView.
I use another vendor's automation product - does this mean I have to convert all my automation to use SA?
No. GDPS provides the capability to coexist with other automation products. This means that you can implement the basic functions that GDPS must have in place and then turn the automation control over to your preexisting automation product. This also gives you a choice to automate any new process with SA for z/OS or your current automation product. It also allows you to migrate your current automated processes to SA for z/OS. This migration can be done over time and does not have to be done all at once.
Can GDPS work with non-IBM disk?
Yes. Although GDPS is based upon IBM remote copy solutions such as Metro Mirror, Global Mirror, and z/OS Global Mirror, these are non-proprietary products. Many non-IBM disk vendors support these products and we have GDPS reference customers running with non-IBM disk. Talk to your vendor to find out which level of the functions they support.
How far apart can I spread my CF's? How far can I stretch my Parallel Sysplex
ISC-3 links can be extended to 100 km without any RPQ. With RPQ 8P2263 using qualified DWDM equipment, this can be extended to significantly longer distances. Distances affect workload response times. More detail on this is described in "GDPS/PPRC 100KM Test" at ZSW02238USEN.PDF (PDF, 216KB).
I need zero RTO/RPO at 1000 kilometers, what can you do for me?
A GDPS/PPRC configuration with a z/OS Parallel Sysplex can be extended to beyond 100 km with RPQ 8P2263. This allows for a Recovery Point Objective (RPO) of zero seconds. By combining GDPS/PPRC with asynchronous solutions such as GDPS/Global Mirror in a cascading remote copy configuration, or GDPS/PPRC with GDPS/XRC in a multi-target remote copy configuration, one can obtain an RPO of zero seconds, with a Recovery Time Objective (RTO) of approximately one hour, the time it takes to IPL an LPAR at the remote site.
If one needs an RTO of zero seconds, then the DB2 Queue Replicator should be considered.
Why can't I just buy the software without the services?
Services are only required on the first license for your location. After you have implemented GDPS in one environment you can buy additional license for that location without buying services. The services are designed to provide your staff with the needed training through skills transfer by working side by side with the IBM GDPS experts. To Implement a high availability and disaster recovery solution such as what GDPS provides will affect all aspects of your IT environment. There are many options that you can customize within the process. The GDPS consulting services will be able to help you pick the correct option that will fit your needs and budget. The services will help you understand the risk associated with each option within your configuration and product selection. IBM wants your implementation to be successful and your staff to be able to maintain this high availability and disaster recovery environment when completed, therefore IBM provides the onsite services to assist and train your staff.
There are many examples of non-GDPS customers running with remote copy who have encountered problems, but didn't know how to handle the situation. A 22 hour outage could have been reduced to one hour or less if they had GDPS.
How hard is this GDPS environment to maintain and how many FTE's are required to run it on an ongoing basis?
This is one of those your mileage will very statements or how long is a string statements. GDPS is a IT infrastructure manager. It automates not only the day to day processes that are required to manage a high availability and disaster recovery solution, it also automates the setup, the monitoring and the recovery of your IT environment. GDPS is designed to reduce the number of FTE’s that would be required to manage this environment without GDPS.
I have an ELA with IBM Software Group, can I roll GDPS into this agreement?
No. GDPS is a IBM services solution. It is not a IBM Program Product. The ELA is from the IBM Software Group within IBM. The GDPS solution has a Control Code component that is a Intellectual Property Services Component ("IPSC"). The IPSC is pre-existing IBM proprietary literary works that is licensed to the end user. IPSCs are not Products or Materials, as such, terms are defined in the IBM Customer Agreement ("ICA"). The terms of the ICA apply to IPSCs as if an IPSC was a "Product". IBM has all right, title, and interest (including ownership of copyright) in IPSCs and IPSCs are licensed, not sold.
How much does it cost?
The cost of GDPS is influenced by many factors. There is the IPSC or GDPS control code license and the consulting services required to implement. The cost of the license depends on which flavor of GDPS is selected. The amount of services that are required will depend on the size of the environment. The type of GDPS and the amount of services required are normally determined in a Technical Consulting Workshop (TCW). The TCW is a fee services offering where IBM will bring the GDPS technical advisors onsite to assess your requirements and educate your staff, at a high level, as to what is required to get from where you are today to where you want to be with GDPS implemented. Each engagement is a custom built solution to meet your requirements. Rather you need an entry level solution such as the GDPS/PPRC HyperSwap Manager or you want a full enterprise wide solution to manage the high availability across multiple platforms and multiple sites we have the capability within the GDPS suite of service solutions to do it. For more information on your direct needs or to schedule a TCW contact your GTS sales representative or send a note to GDPS@us.ibm.com
How long does it take to install?
There are several factors that determine the amount of time to implement a GDPS solution into a clients environment. The solution to be implemented. Then we need to look at the size of the environment and what other projects the client has to implement in the same time frame. Will all the disk and or tape be upgraded. Will the processors be upgraded. Is the fiber in place between the sites. Do we need to implement remote copy. The list goes on. To determine the amount of time we need to understand what needs to be done. Where you are today and where you want to be tomorrow.
Typical numbers can be:
1-3 months for HM
3-6 months for GDPS/XRC or GDPS/GM
9-12 months for GDPS/PPRC
Now that we have Basic HyperSwap, why do I need GDPS?
GDPS/PPRC HyperSwap Manager has some significant advantages over basic HyperSwap.
GDPS/PPRC HM will
- manage all vendor’s disk that support the IBM Metro Mirror.
- provide a 2 site disaster recovery capability
- provide consistency across all target volumes when the Swap fails to complete
- provide a migration path to full GDPS
- provide the high availability in the metro area when combined with GDPS/XRC to provide the 3 site GDPS/MzGM solution
- provide the high availability in the metro area when combined with GDPS/Global Mirror to provide the 3 site GDPS/MGM solution