02-08-2011 09:31 AM
https://app.ti.com/path1/file.html goes to Web servers A & B -> "
02-08-2011 01:48 PM
Hi,
for SSL communication is it NOT possible, because CSS can't see application layer conversation (HTTP protocol).
Answer to second part of your question is yes, it is possible. CSS (in general load-balancer - ACE, CSM, CSS, proxy server, etc) can see application layer communication and regarding information on this layer can direct communication to specific destination server(s).
Configure two contents with different destination servers (services) - in your case C,D and A,B. For each content define differnet 'url matches'. You have two possibilites for 'url matches':
url [“/url_name”|“/url_path/*” [eql eql_name|dql dql_name {eql_name}]|urql urql_name]
example1:
(in this case you can use regexp)
content MyContent1
url "/path1/*"
! or url "//domain/path1/*"
add service A
add service B
etc
content MyContent2
url "/path2/*"
! or url "//domain/path2/*"
add service C
add service D
etc
example2:
(for urlq is not possible to use regexp!)
urlq Path1
domain "domain_or_ip"
url 10 url "/path1/path1.html"
url 20 url "/path1/path2.html"
urlq Path2
domain "domain_or_ip"
url 10 url "/path2/path1.html"
url 20 url "/path2/path2.html"
content MyContent1
url urlq Path1
add service A
add service B
etc
content MyContent2
url urlq Path2
add service C
add service D
etc
For more information see documentation. Hope it helps.
--
martin
02-09-2011 01:57 PM
Hi Martin,
Wow you are a CSS Expert, thank you for your reply.
I have a few more questions:
1) For example 1:
I have to use url "//domain/path1/*" & url "//domain/path2/*" if I don't use domain name (only VIP address) in a content rule, correct?
2) For example 2:
urlq Path1
domain "domain_or_ip"
url 10 url "/path1/path1.html"
url 20 url "/path1/path2.html" -> this line shoud not be included ?
urlq Path2
domain "domain_or_ip"
url 10 url "/path2/path1.html" -> same here, this line shoud not be included ?
url 20 url "/path2/path2.html"
Both example configs will work, just a matter of preference, right?
Thanks again,
Diane Ly
TI Network Infrastructure
02-10-2011 01:19 AM
Hi,
at 1),
it depends how is web service (VIP) requested by clients. if clients use http://domain/, you have to use url "//domain/..", if there is IP, vice versa. My recommendation is - is it possible define url "..." without domain, only with path (url "/path1/*"). in this case, content is matched for both - dns name and ip address
at 2),
count of url definitions depends on you, this is only example. but remember, as I wrote, for urlq is not possible use regexp! (see documentation)
Hard to say, which one is recommended. It depends on url definition. For urlq definition url paths is not possible define url path using regexs, url path configured under content does. In general, regexps (depends on definition) use more memmory resources (for example: *something* use more memmory than someting*).
My recommendation is: use what is better for your web service. If you need define two contents for /path1/... and /path2/... with lot of subpaths, use regexps /path1/*, /path2/*. If your count of subpaths is limited, use urlq with exact url paths definitions. It's up to you.
It is helpful?
--
martin
02-10-2011 01:18 PM
Hi Martin,
1) Clients are using http://domain/, thus I need to define url "//domain/.." as you have stated.
2) I'm still waiting for the "actual" URL paths from the application team in order to decide which method to use - Regexp vs URLQ definition - to begin testing.
I know I have limited memory resource (as listed below) & will try not to make it complicated:
System Resources for CSS501-SCM-INT:
Installed Memory: 268,435,456 (256 MB)
Free Memory: 135,414,448 (129 MB) ****
CPU: 0% (5Sec) 1% (1Min) 0% (5Min)
Buffer Statistics:
Buffer Pool: 0
Size:2048 Total:3072 Available:2792 Failures: 0 Low Buffer Count: 2748
Buffer Pool: 1
Size:2048 Total:3072 Available:2800 Failures: 0 Low Buffer Count: 2800
Buffer Pool: 2
Size:2048 Total:2048 Available:1956 Failures: 0 Low Buffer Count: 1900
Thanks Martin, you've been a big help!
Diane Ly
02-11-2011 12:47 AM
regexp is not scarecrow, it's ok use 'normal' regexp, i.e. '/path1/*'.
good luck with testing
--
martin
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide