Thanks for the suggestion. I've posted in the public TS newsgroup.
M@ On 7/8/06, Susan Bradley, CPA aka Ebitz - SBS Rocks [MVP] <[EMAIL PROTECTED]> wrote:
This sounds like a question for a MSDN/TS list/newsgroup with some monitoring tools thrown in as you do your tests. I can tell you that in our little networks, things like smb signing enabled on our DCs add about a 20 to 40 percent overhead to file transfers and apps (ergo one of the reasons we're a bit insane to be making our DCs file servers). We've also seen speed issues affected by NIC drivers....and the selection of a static speed versus auto-sense on the nic. Just reading that laundry list of what that app is having to go through.. each possibly needing a little tweak here or there...sounds to me that a test, perf mon and other such monitoring is needed to determine if he's right? Matheesha Weerasinghe wrote: > Basically the reason I am inquiring this is because of performance > issues which were blamed on application redirection. The appdata was > on a cluster in this particular instance. Siting the fact that there > are more components involved in the data path when appdata is accessed > from a cluster , the PSS guy basically didnt personally seem to > approve the design. And it seems like quite a few guys share his > opinion. As he explained, in a normal file server the client will go > through the file server's nic, the ide/scsi controller and then to the > disk(s). In a cluster environment, the client goes through the cluster > node's nic, the node's HBA, fibre switch/hub, SAN controller, and > finally disk(s). And in the case of small files the SAN was not very > performant especially with big volumes with lots of files. > > In the webcast I mentioned in the original email, in slide 22 of the > presentation available at > http://www.microsoft.com/emea/itsshowtime/sessionh.aspx?videoid=26 for > group policy tips and tricks Mark Cribben recommends against it. I > would say the main reason for that recommendation is network latency. > > We are designing some file servers at the moment for the client and we > have some design considerations and fears. Basically we are wondering > whether to do away with appdata redirection altogether and leave it in > the profile itself. One of the suggestions is that we may take a hit > in logon time to download profiles , but app performance will be good > as the files are cached locally during the TS session. > > We would like to use appdata redirection if at all possible. But we > dont want to sacrifice app performance for it. i.e. We dont want to > wait too long while the app is looking for ini files etc.. > > Thoughts? > > M@ > > On 7/8/06, Susan Bradley <[EMAIL PROTECTED]> wrote: > >> >> Sorry read the original post and saw it was specifically about TS. >> >> TS is one of those things that if the application loves the TS >> environment, >> I don't think we've seen too many issues... and that's usually the >> key... >> there are some applications that just don't work well and the vendor >> states >> so in a TS/Citrix setup and would have problems redirecting. >> >> I know that we redirect 'normal' stuff like My Docs folder all the >> time over >> a TS... but apps like Word and Excel don't have to maintain a constant >> connection to a data file. >> >> >> Susan Bradley <[EMAIL PROTECTED]> wrote: >> Please correct me if I'm wrong.. but in the era of Howard/LeBlanc and >> Howard/Lipner's Secure Coding and SDL books.... currently written >> software >> from Microsoft is indeed following their "best practice" guidelines. >> >> (Which my only complaint wtih both books is that they are paperback >> and not >> hardbound and thusly when I throw them at crappy app developers like ... >> oh.. say.. I don't know....Intuit... the bruise on the head of the >> Dev folks >> there will be slightly lessened.... the SDL book so far is very >> interesting....) >> >> Older software that they purchased .. granted that statement cannot be >> made... >> >> And isn't your situation solvable with having on your patch test >> matrix a >> check box that says "ensure app data redirect is still functional"... >> and of >> course testing that patch before it's globally deployed? >> >> Matt Hargraves <[EMAIL PROTECTED]> wrote: >> I believe the reason they recommend against this is because all >> applications >> are different. Another problem is that there is no guarantee that the >> application will remain the same. Patches and updates can change >> more than >> just a file here and a file there, they can change settings such as >> these >> and trying to redirect the location for that data can end up with a >> situation where the application during an update is trying to pull your >> information from %userroot%\appname and it's really at a completely >> different location. >> >> If all application vendors use MS best practices for programming, it >> would >> be great, but unfortunately not even MS always uses their own best >> practices. >> >> Redirecting application data can work fine for months or even years, but >> then you get an update to an application and *bam* everything's >> broken and >> you don't really know why and you spend days (or worse, weeks) trying to >> figure out why everyone's broken and realize that your problem is >> that the >> application data is being redirected and that's the source of the >> problem. >> >> Matt >> >> >> > List info : http://www.activedir.org/List.aspx > List FAQ : http://www.activedir.org/ListFAQ.aspx > List archive: http://www.activedir.org/ml/threads.aspx > List info : http://www.activedir.org/List.aspx List FAQ : http://www.activedir.org/ListFAQ.aspx List archive: http://www.activedir.org/ml/threads.aspx
List info : http://www.activedir.org/List.aspx List FAQ : http://www.activedir.org/ListFAQ.aspx List archive: http://www.activedir.org/ml/threads.aspx
