Could you provide more details about the Architecture?
-What is the Internet connection upload speed in mbps of each site?
-Are you replicating the Archive Server, or accessing remotely?
We have a 4mb line from here to the states, although we are looking at upgrading to 10mb min.
In terms of replication, we are not currently replicating our archive server back to the states etc.
Access is not via VPN etc. We have direct visibility to the servers over our network
The initial set up was done by an employee who has left, so I'm picking this one up.
David, I believe this more of network issue than a EPDM one. If your sites have capability to host a Replicated Server, I would strongly advise to do so. Replication is design for this scenario and it works well. If you must continue to "WAN" into your main location I would definitely go to a 10mbps connection.
Sorry perhaps I wasn't clear
The replication is pulling files to our local archive server, as they are called from any other sites.
As it currently stands we have the speed issue when looking at files created on the local uk archive server. So it's effectively looking for files at the same site therefore bandwidth of files shouldn't technically be an issue in this example.
Our VAR has said that the performance is down to the latency of connecting to the sql. With the main db being queried to find file locations etc.
For me this doesn't seem right. If epdm truely is an enterprise product, then how do other companies with multi sites get past this. We are sending a small ammount if data in relation to the query which I can't believe would require 30 sec plus response time.
The whole point of archive servers is to get round the issue of bandwidth and lag, but if the queries still take that long then any benefits are lost.
Personally I believe that our systems are not set up correctly/there is a bottleneck somewhere so I'm interested in seeing if anyone else has a working system across continents with similar latencies.
Have you got multiple site installs?
We use EPDM across many sites globally. While users always want it to be faster, it is tolerable. Our SQL server in in Tampa, FL and we have archive servers throughout the world. Every operation must go through the SQL, even a right mouse click to add a fillet to a face in SolidWorks pings the SQL server. There are many things that you can to to minimize the amount of pings to the the SQL server, like turning off the Add-In for EPDM in SolidWorks (in 2015 this is not required). Upgrading the bandwidth should help, 4 Mbps is on the low side I would say. Our sites range from 10-40 Mbps for bandwidth. Below is a table of the latency to each of our sites we have connected. Anything over 150 is considered by SolidWorks corporate to be high and could cause less than ideal usage. But we have sites that are well higher that operate quite well. 5-10 second delays sometimes occur when clicking on files in Explorer during high usage we have found. We have 300 client installations and about 125 on at any giving moment globally right now. A dedicated VPN or MPLS connection should be in place.
Are your servers virtual or physical? Ours are all virtual. We are also running SQL 2008R2. We have 32 GB of RAM on the SQL server and set SQL to only use 26 of it leaving the rest for the OS so it was not starved for RAM.
I took these other actions over the past several months to speed up our database as it was not as snappy as we wanted:
Made SQL server and Archive server separate at main data center site Changed default db from master to conisiomaster. See: Re: Slow EPDM Performance - remote clients using Replicated Archivers Rebuild/Reorg database maint. plan every Sat at 1am. See: FISHER/UNITECH Blog: Maintaining a Healthy SQL Server for EPDM Defragged SQL Server and set a monthly plan to do so 1st of the month at 1am Changed SQL recovery mode from Full to Simple on production Vaults. See: Set Transaction Log for EPDM vaults to Simple Recovery Model - YouTube Shrink database for Transaction logs Autogrowth was different between all Vaults (We have 5 Vaults active, I know it's crazy, we are working towards combining them), (1mb to 100mb) so set all to 100mb. And set to size growth 64mb vs 10% growth
Make sure you work with a good SQL expert when making any of the above changes.
Also, depending upon the actions that are occurring in a workflow, the time it takes to approve documents can go up dramatically. However consider the amount of automation that EPDM provides by going paperless. Even if a click takes 5-10 seconds, there are savings in other areas. But 30+ seconds is way too long to be tolerable.
Continue to work this out with your VAR and SolidWorks.
Michael, thanks for that detailed feedback. Its very useful to have this information, including a breakdown of the latency.
Could you clarify whether all of the sites are seeing a "huge delay in trying to perform any actions" or does your observation only apply to the UK installation? Also what is the latency between the "US" and "CA" (Canada?)?