Single session max throughput on 5555-X

We have some ASA 5555-X firewalls running 9.1(2) with the IDS modules installed that have a pretty basic configuration used to separate multiple internal networks. The firewalls are able to pass over a Gbps of traffic pretty easily as long as it's sourced from multiple services but if we try to do a high speed transfer such as backup to SAN running over a single UDP or TCP session the traffic seems to peg out around 400Mbps. We can verify that the server and SAN are capable of much higher transfer speeds by putting both on the same side of the firewall so I'm convinced the firewall itself is the limiting factor here.
I was wondering if anyone has come across this before and/or has any suggestions for how to boost the speed. I am really trying to avoid something like bridging the networks with multiple interfaces or route exporting VRFs but our DBA's are complaining nonstop about backup times in those environments. There are no Service policy QOS settings set on those interfaces and are NAT exempt so the firewall is not doing anything what I would call special.
Thanks,
Richard Hillius

Hello Richard,
At this point it would be good you do some captures on both interfaces of the ASA in place for this traffic and then see the round trip time just to make sure the ASA is the one generating the slowness on the network.
And agree the ASA can do QoS for specific traffic (so for all traffic that you defined, not just for a single session).
Can you share the following commands:
show interface | include errrors
show cpu usage
I could start providing you comamnds but I think it's better to give you the link :
http://www.cisco.com/en/US/products/ps6120/products_tech_note09186a0080ba9521.shtml
Here is what you need to undertstand about the ASA, let us know the test you performa afterwards.
For more information about Core and Security Networking follow my website at http://laguiadelnetworking.
Any question contact me at [email protected]
Cheers,
Julio Carvajal Segura

Similar Messages

Max no# of threads in a single session

Hi all,
I have a scenario wherein I have connected to SAP from JAVA using XBP interfaces.
I want to call BAPI_JOB_DEFINITION_GET concurrently in a single session using sync threads. The problem what I am facing here is, I am able to handle 15 to 20 connections but for high volumes of threads(50) the JVM is crashing.
I guess there is some limit set from SAP end.
Could anybody provide any suggestions or advice about, the maximum number of connections and how can I set the parameter to modify the limit.
Thanx in advance
Milind.

This topic should be posted in the Java Programming section, not the xMII section...

Max no# fo connections in a single session

Hi all,
I have a scenario wherein I have connected to SAP from JAVA using XBP itenrfaces.
I want to call BAPI_JOB_DEFINITION_GET concurrently in a single session using sync threads. The problem what I am facing here is, I am able to handle 15 to 20 connections but for high volumes of threads(50) the JVM is crashing.
I guess there is some limit set from SAP end.
Could anybody provide any suggestions or advice about, the maximum number of connections and how can I set the parameter to modify the limit.
Thanx in advance
Milind.

Hi
You will have to do memory settings for the number of threads you are trying to access.
Procedure:
Go to configtool and try to change the heap size.If you can tell the configuration of your OS i can help you in setting the corrcet heap size and JVM parameter.
\usr\sap\SID\JCXX\j2ee\configtool and run configtool.bat
1.Go to instance----
Check the heap size
2.Expand the instance node
Click the dispatcher node----
Check the heap size
Click the server node----
Check the heap size
Change the values.If you have confusion send the present values to me.
Message was edited by:
rajat anand

How to do Multithreading in Oracle PlSql using a Single Session

Hi All,
I have a complex but interesting requirement.
1) I have a table with 90k rows.
2) My PlSql Package picks up all the 90k rows one by one, does some processing/transformation on it
3) if the row satisfies some business rules, then it gets inserted into an application table. If the row is not eligible, then it inserts the row into a Error table.
4) After processing all the 90k records, If my Error Table Count reaches 10% of the input count (90k), then all inserts (i.e. both application table inserts and
error table inserts) must be rolled back.
5) Right now, we are achieving this logic in a single session but there are far many performance issues, because processing has to go through row by row.
Is it possible to implement the above logic as "multi threading within PlSql" whereby i can run Step 2 in 10 parallel slaves and if the total error count of all the 10 slaves exceeds 10% of the input data, then there should be a rollback for all 10 slaves.
+A sample testcase is given below. Issue is: rollback statement in p1 is not taking any effect (because plsql spawns sessions to execute p2 which autocommits itself on graceful exit). But is there any other way by which i can make the rollback to be effective+
create table t1 (a number);
create table j1(b number);
create or replace procedure p1
as
l_cnt number;
i number;
l_job number;
begin
     for i in 1 .. 5 loop
        dbms_job.submit( l_job, 'p2('||i||');' );
    end loop;
    commit;
    loop
        dbms_lock.sleep(30);
        select count(*) into l_cnt from j1;
        exit when (l_cnt = 5);
    end loop;
   rollback;
end;
create or replace procedure p2(i number)
as
begin
insert into t1 values(i);
insert into j1 values(1);
end;
/

Basically your current 'algorithm' can be described as 'slow-by-slow processing' (co Tom Kyte)
Relational databases are about sets not about records.
Consequently your approach must be qualified as evil.
Instead of fixing the problem (and getting rid of slow-by-slow processing), you propose to add even more evil by adding multi-threading to the mix.
The issue you describe is not an issue at all, it is just making clear you are heading for disaster.
An approach that works quite nicely is
- add an extra status column to the table to be processed.
- UPDATE the status column of all records that do not qualify. By UPDATE I mean the UPDATE statement, not record by record processing
- (BULK)-INSERT the remaining records, for which the flag is NULL.
The simple guidelines for application development are:
Only do it in PL/SQL when you can't do it in straight SQL
Only do it in Java when you can't do it in PL/SQL
Only do it in 3GL when you can't do it in Java.
Your 'strategy' is doomed to fail.
Get rid of it.
Before it is too late, and your heart is pumping like mad, because your problems are increasing, and increasing, and increasing, and you have to work 7 x 24 to deal with 'performance problems'
You have already demonstrated your approach doesn't work.
Don't try to make it worse.
Sybrand Bakker
Senior Oracle DBA
Experts: those who do read documentation

Can I download a file in different sessions or should it be in a single session? In other words can I pause the download of a file, shutdown/restart my system and then resume download from where I left off? Or will it start from the very beginning?

Can I download a file in different sessions or should it be in a single session? In other words can I pause the download of a file, shutdown/restart my system and then resume download from where I left off? Or will it start from the very beginning?

If a server supports resuming downloads then the download should continue where it left off.<br />
You can test that by simple starting a large download and wait a minute or so before pausing it.<br />
Then you can close and restart Firefox and see if resuming works.
You can set the pref browser.download.manager.quitBehavior to 2 on the about:config page.<br />
That will make Firefox ask for confirmation if you want to cancel the download if you close Firefox.
See:
* http://kb.mozillazine.org/browser.download.manager.quitBehavior

Message Driven Beans: how to receive and send in a single session?

Hello
How to receive and send JMS messages in a message driven bean using a single session?
Thanks.

Are you referring to writing code within the onMessage method itself to receive JMS messages?
If so, that is not a pattern that is recommended, especially not in the case of a blocking JMS
receieve. If you must write an explicit JMS receive within an EJB component it's best to use
a "check the queue" approach where you give a very small timeout for the receive. That way
if the message is there it will be returned but if not the call will not block.
You'll also need to take the transactional settings of the onMessage method into account.
If your are sending a message and then waiting for a response, those two actions can
not be done within the same global transaction. For that, you'll need to either use
BMT and perform each in its own transaction or just not use global transactions at all,
in which case each send and receive operation will be atomic.
--ken

Single session for all Application Modules

Hello,
I've got an application that has several Application Modules. All these modules need some data that are set from time to time. I've written some code that calls getSession().getUserData().put("myData",myData). After this method was called in another Application Module i want to retrieve myData with getSession().getUserData().get("myData"). But the result depends if I use 2-tier or 3-tier. In 2-tie I get myData while in 3-tier I get null. My question is:
Is there a Session object or any other data instance that could be accessed from all Application Modules for on application instance.
E.g. I've got two user (A,B) who start my application. Through using my app several instance of Application Modules are created. All Application Modules that are created for the instance of User A should be able to share his user name and some other dynamic data while the Application Modules created for B should have no access to that data.
Doe anybody know if static data of an Application Module are a possible solution?
regards
Joerg

Hi Joerg,
I am not 100% sure what you are trying to achieve, and with which version of JDeveloper/ ADF, technologies (JSP, Swing), etc, but I can tell you that I have been experimenting with single session accounts myself within a rich-client Swing environment, as I have experienced Oracle DBA guys who want to control user authentication via Oracle accounts (I have achieved this via the JCLoginDialog mechanism, and a bit of a hack)... Frank Nemphius (sp?) posted something about a white paper coming out on a related subject but I don't know where he is with that.
Anyway, I managed to do it in a prototype by nesting several application modules within a Root Application Module that was essentially simply a Container for the other application modules. Using this mechanism I was able to bind my root application module to any panel I wanted and it would inherit the roots Session.
ie. I used many application modules and it only used one Oracle Session.
However, my reservations -
1) This solution was not tested robustly!
2) I doubt this would be ideal for a distributed app with thousands of users - My solution is for a very complex app, with a limited number of users.
Finally, I cannot confirm this is a great solution - I can only tell you it worked in my prototype, and I am currently awaiting the ADF source - when my company and Oracle can decide what level of support we currently have and need!?!? - what exactly does being an "Oracle partner" mean??? Its all very boring and un-interesting to a Java person like me... ;-)
Once I have the source I will look at how it actually works and perhaps be able to answer questions like this more definitively. Hmm - I wonder if I could get the Oracle JDBC driver source too?? .... Is it written in C or C++?? I know them too... ;-)
Cheers........Dean

RDS 2012 R2 - Allow Some Users Multiple Logon Sessions and Restrict Other Users to a Single Session

In Server 2012 R2 RDS, is there a way to allow some users to log on multiple times, but restrict other users to a single logon? On an OU or AD group basis?
I know there is a GPO setting under Computer Configuration for restricting users to a single logon, but this does not allow me to differentiate on a user basis (only on a per server basis).
Thanks,
James

Hi James,
From my perspective and knowledge, sorry to say but there is no such option\way to provide this permission at same time. If a user specifies a different program to start when the user connects to the RD Session Host server, a new session will be created on
the RD Session Host server for the user, even if the RD Session Host server is configured to restrict users to a single session. A user can specify a program to start on connection on the Programs tab under Options in Remote Desktop Connection.
Hope it helps!
Thanks.
Dharmesh Solanki
Please remember to mark the replies as answers if they help and unmark them if they provide no help. If you have feedback for TechNet Support, contact [email protected]

B2c single session trace error

hi,
when i click on 'single session trace' in b2c/start.jsp page,
it ended with error http code 400.
but when i clock on the webshop ID, it is accessible.
however, when i decided to register a user, it ended up server not found error.
May i know is there any solution or section where shall i look for?
thanks
Shanti

hi, does anyone faced this problem before and found the resolution?

BT Business Hub 3 - What's the max throughput figu...

Hi all, I have Infinity 2, using the "BT Business Hub 3" - can anyone tell me what the max throughput figure is for this device? I'm thinking of replacing it with a more secure option and I'd like to be able to compare the numbers. Thanks, RS

In what reguard?
The wan port is 1gbps.
Three of the lan ports are 100mbps and one is 1gbps.

Sharing a single session among multiple portlets for single application (e.g. wInsight)

Does anyone know the best practices way for sharing a single Session for an application (running on a single Porlet server) among multiple portlets on a page?
For example, the application wInsight uses 3 frames (navigation, search results - tabular, search results - chart) which need to communicate via the application's session scope. It is not just that we do not want for a single user to be running multiple sessions, but they really do share variable parameters this way.
So far I can only think of trying to build a one true portlet and two empty containers on page load. I would use the portlet to establish the session with the Portlet server and then transfer the session information to the other two empty containers to instantiate their respective portlets. I haven't tried this yet, but will do so soon...
Any help and feedback would be greatly appreciated.

Karen,
You can easiliy create several portlets out of a Struts application. For detailed instructions, refer to the Struts section of the Portal Developer's Guide.
Good luck,
Peter

Restrict RMAN to start single session per database

Hi,
Is there way we can restrict RMAN to start single session per database. we are using RMAN for full database Backup.
I can not restrict user session as same catalog user is shared with other database as well.
ORACLE : 11g

Ok :)
Backup files created by RMAN must be tagged with the local system name, and with RESTORE operations that tag must be used to restrict RMAN from selecting backups taken on the same host. In other words, the BACKUP command must use the TAG node name option when creating backups; the RESTORE command must use the FROM TAG node name option; and the RECOVER command must use FROM TAG node name ARCHIVELOG TAG node name option.
RESTORE DATABASE FROM TAG '<node name>'
RECOVER DATABASE FROM TAG '<node name>' ARCHIVELOG TAG '<node name>'
http://download.oracle.com/docs/cd/B14117_01/server.101/b10823/manage_ps.htm

NI MAX throughput error and Camera attribute out of range

I recently purchased a Point Grey Grasshopper3 camera, it seems to work fine in the Fly Cap software point grey provides, but I get an error in NI MAX when I try to do a simple grab (shown below). This error shows up every time I try to grab an image the first time after opening NI MAX, even though throughput is set to the max. Then I change the throughput max under device control to any number around the max of 125000000, and the camera works fine for MAX. It will grab images as much as I need. Then I save the camera, exit MAX, open MAX again, it shows my saved settings, I go to grab images and I get the error again. So I change the range to whatever number and it works again in the same session in MAX.
I did have a point grey Zebra2 working fine, and was using it in my LabVIEW program with IMAQdx to acquire images. When I try to use grasshopper3, LabVIEW returns an error of camera attribute out of range from the camera configuration VI. The VI's I am using to test the camera is just open camera, configure acquisition, start acquisition, get image, unconfigure acquisition, and close camera from IMAQdx. I did twice open MAX, change throughput, save camera, close MAX without grabbing an image, and then was able to acquire images in LabVIEW on the first run. Then I would run the VI again and it would return the error for camera attribute out of range. I have changed the packet size, resolution, and frame rate of the camera with out noticing a difference in the errors.
Since I am able to stream images from the camera in point greys fly capture, and in MAX after toggling the throughput setting, it leads me to believe there is some sort of MAX/LabVIEW error preventing me from capturing the images.
Any ideas? Has anyone else had this issue with MAX and LV?
Solved!
Go to Solution.
Attachments:
MAX error.png ‏17 KB

It looks like Point Grey cameras have shown this error before based on out of date firmware. See this link for one such instance. I would monitor the camera file (*.icd) located in the IMAQdx data folder in Public Documents to see if the actual value is being overwritten after the camera configuration is saved in MAX. There is a chance that the GigE .XML file that is transferred from the camera is overwriting these values to the unacceptable value. See this link for a KB regarding this error, which leads me to believe this may be the issue.
Karl G.
Applications Engineer
ni.com/support

Single Session per Authentication/MAC

Hi All
We are in process to deploy a wireless for a customer with ACS, where we want A single User/machine to have a login checked with External Identity store and have only one session at a time.
i.e. if User A logged in with Machine A, he should not be able to use Machine B for the same authentication even if the Machine B is having MAC authenticated, (please note that MAC Authentication is not necessory but one user should use only one machine)
I am a little new to the ACS/Wireless, any help would be highly appriciated.
Many thanks for reading me.

Hi tarun,
I think you are looking for the new feature in ACS 5.3:
http://www.cisco.com/en/US/docs/net_mgmt/cisco_secure_access_control_system/5.3/release/notes/acs_53_rn.html#wp195861
Maximum user sessions
Allows you to restrict the user from too many concurrent user sessions. The permitted number of concurrent user sessions is between 1 and 65535.
For more information on this see:
http://www.cisco.com/en/US/docs/net_mgmt/cisco_secure_access_control_system/5.3/user/guide/access_policies.html#wp1176806
Hope this help

FAILED_OVER for single sessions. No VIP failover

Hi all,
On a 10.2.0.3.0 2-nodes RAC (AIX 5L) I am seeing from time to time single of few sessions marked in gv$session with FAILED_OVER='YES', but no services failover occurred. Those sessions are still connected to the preferred instance, that is in their tnsnames.ora they refer to a service running, let's say, on instance 2 and they ARE running on instance 2.
I am wondering if this can be due to some kind of connectivity issue, that is the client doesn't ping the VIP address anymore (and vice-versa) and then the session is marked as FAILED_OVER.
Has anyone seen something similar?
Thanks for any feedback,
Riccardo

With Oracle 10g you do connection time failover with setting in clients tnsnames.ora file all nodes addresses: Check following link:
http://docs.oracle.com/cd/B19306_01/network.102/b14213/tnsnames.htm#i477297
If your clients tnsnames.ora file is ok, Then client tries to connect addresses one by one. And this way it does not matter even if some server of the cluster is down. But of course when you shutdown one database instance it's connections will be dropped. Althought you can have SELECT clause failovers with Transparent Application Failover (TAF).
So you can run rolling update without shutting down whole RAC database as long as your clients tnsnames.ora is configured correctly. But those dropped connections need to be handled in application level.

Single session max throughput on 5555-X

Similar Messages

Maybe you are looking for