Disaster Recovery stratergy for ERP 8.0 on AS/400

Hi List,

Is there any one out there who has implemented a Disaster recovery stratergy for Xe/ERP 8.0 on AS/400

Would appreciate some suggestions on the basic approach that I can follow.

Should install JDE on a new set of servers(I know theres a catch to the licensing part) , and then keep them in sync..by restoring backups daily etc..

How would my users switch to backup servers incase of failure


Or should I just mirror my servers(through journalling i presume..but would this affect my perfromance) on to backup servers.
Again in this approach how would i switch over. Rename and reconfigure TCP/IP on my backup servers..?




Or should I go for a clustering kind of setup.

Would appreciate any help in this area..as I am quite confused about the whole thing as you can make out from my post any way..lol



Thanks in advance

Icecube,

I am no means an expert on the subject, but here are a few thoughts.

Disaster recovery plans are designed to cover a catastrophic loss due to fire, flood, earthquake, etc… Therefore, while it may be a good thing to have a server cluster, if the cluster is in the same building (or possibly even in the same state as some disasters will encompass an entire state, such as an earthquake or hurricane), it really does not protect the company in case of a catastrophic loss, like the entire building burning to the ground. Your plan should cover this type of contingency. If your company has multiple locations located throughout the country, setting up a duplicate server room that is kept in sync or would allow the restoration of your backups would be a good start. If your company does not have multi locations, you may wish to contract with a disaster recovery center which will guarantee availability of compatible hardware within a certain number of hours.

Your plan should also encompass what computer assets need to be restored and how quickly. For example payroll functions are usually your #1 priority. So if you run payroll in JDE, you will need JDE up and running, a way for the payroll department to access JDE which could be a fat client at the recovery center, a method of getting employee’s time into the system and a method of printing checks. This could mean the restoration of two servers in addition to JDE. For example a server for a web based time entry system and a server for a custom check printing program such as Optio. So don’t limit your focus to just JDE.

Also, don’t forget your deployment server. I was at a client who had not backed up their deployment server since “go live”. I asked them, how will you build a new server package after a disaster without your deployment server? How will you restore all of the custom objects that have been created? Horrified looks were the response.

Good luck.

Hi Smith,

Thanks for your response. Well to start with what I am actually looking at is a fail over setup. I dint mean Disaster recovery in the actual sense. If my main server goes down for some reason i need an alternate server/setup to take over. A couple of hours down time for the switch over is acceptable.

We run HPUX but I would assume IBM would offer the same functionality - since your disk array is effectively redundant and probably has a listed uptime of over 99.9999% then all you need to do is actually set up the second server as a passive failover solution hooked up to the same disk array. The HP solution basically says that your main server has a real name of something like server1 and a "package name" of pkg1. When the heartbeat connection between server1 and server 2 fails then server 2 makes sure that the disk array is unmounted by server1 and then it mounts the disk as pkg1 and thus assumes the role of the functioning server. One World is then configured to look for the "package" name rather than the real server name and everything works like a champ.

And if your RAID controller dies...?

We have a fail-over or High Availability AS/400 in another data center located 25 miles from our main data center. We use a product called Vision Suite from Vision Solutions. We do remote journaling of all the JD Edwards libraries as well as the IFS objects for JD Edwards. We see no performance issues at all nor is there a licensing issue as the license is on the deployment server. We have configured a virtual IP address that all external servers and users access that floats between the AS/400s and is only active on the one that is the primary at the time. We have successfully done what is called a role swap and our target or fail-over box became our primary and JD Edwards fired up and we were able to run jobs as usual. If you need more information please feel free to contact me.

Well our HPUX boxes run multiple raid controllers so if 1 dies then the other takes over. In the event that all the raid controlers went out (or say the back plane supporting them) then the servers would fail over just like they should in the event of any other kind of hardware failure. Was there something in particular you looking to point out or solve? Hardware failures are hardware failures - doesn't matter what the piece of hardware is. I was just throwing out what works for us. We also have a true disaster recovery config that has a hot site in Philly but that didn't seem to be the point of the initial inquiry.

Relax...

Many times people focus on drive parity and forget the controller (it sure leaves a puzzled look on their face when a controller goes bad). You hadn't mentioned redundancy in the controller so I was throwing it out there for you.

I'm glad to see you have it covered...