Hello List, I am sorry for keeping silence again, just changed my employer 2 months ago. Now I am with Pythian. :) Thank you everyone who contributed and will contribute to Oracle Apps Community and to that thread in particular. Talking about shared APPL_TOP I do think that there is no universal answer there. Each organization have it own infrastructure, best practices (read internal knowledge and experience) and infrastructure development strategy. Not so long ago I was given a dream task to design an Infrastructure from the a white page for existing EBS 11i environment with 3k users using that system on daily basis (we migrated that system from legacy HW to a new one). As Shared APPL_TOP we used two dedicated servers (we called them Infrastructure servers) formatting RedHat Active/Passive cluster (We ran CM on those to hosts in addition to every thing else like OID and BPEL). Those two servers provided NFS service to all other nodes (we did have 2 Internal Web and 2 External Web servers). The storage (volume) was coming from the same SAN there DB files were located (if SAN goes down, DB goes down and there is now use of NFS services anyway). During 2 years this configuration worked almost perfectly. We have lost one of the Infrastructure hosts several times. Other was taking over NFS service transparently to others Apps servers (NFS Clients). NFS & RH Clustering is included in an Red Hat EE distribution (starting from version 5). The only improvement what I would make is to run a separate NFS cluster in between of each pair of Apps servers (Internal Apps, External Apps and CM-s). As it would speed up patching exercise as in our case adpatch too significantly longer for Internal/External web tiers. As I said the end solution is dependent on many factors unique for an organization. The only thing that I would probably suggest is not leaving NFS running on one server only. Try to ensure High Availability for Sharing File system at least on the same level your rest of infrastructure have (e.g. DB, network etc). Few other comments are as following: - Do not use OCFS (1/2) those IMHO will not be developed anymore (remember ACFS is now released). - For oother customer we run NFS on DB servers as DB had Cluster services installed for RAC implementation anyway. - There are so many different solutions for ensuring FS availability (e.g. Net fillers etc). Choose an appropriate one for your case :) If you do have any questions please do not hesitate to ask, Yury PS I wouldn't suggest to use ACFS without good amount of testing at this stage (to new product). If you GoLive ETA is in 6-12 months I woud start to consider it. On Thu, Oct 29, 2009 at 4:32 AM, biz arrays <bizarrays@xxxxxxxxxxxxx> wrote: > Hi Yury, > > Quick question on Shared APPL_TOP. We have test nfs no problem with, because > it is a Single point of failure, are there any other certified options that > could be used. > > I know few people use ocfs2 which is not supported for Appl_top file system > sharing. > > Are you aware of any better ways of configuring shared APPL_TOP. (any > product(s) from storage vendors like EMC) > > On Sat, Jun 14, 2008 at 9:55 AM, Jurijs Velikanovs <j.velikanovs@xxxxxxxxx> > wrote: >> >> If there are no exceptions rules defined, then first which will check >> queue. >> J. >> >> On Tue, Jun 10, 2008 at 10:53 AM, khabeer khayyam <khabeer1983@xxxxxxxxx> >> wrote: >> > Hello Everybody! >> > >> > If 2 Standard Managers are configured, then which one will >> > pick the request when the request is made. >> > >> > >> > Thnks n Rgds, >> > Khayyam >> > >> > >> > ________________________________ >> > Bring your gang together. Do your thing. Find your favourite Yahoo! >> > Group. >> >> >> >> -- >> Yury >> +371 29268222 (+2 GMT) >> ============================================ >> http://otn.oracle.com/ocm/jvelikanovs.html >> > > -- Yury Oracle Certified Master Mobile number: +61 405728926 ============================================ http://otn.oracle.com/ocm/jvelikanovs.html