Shared SO's and Load Balancing

On Wed, 21 Jan 1998 17:24:33 -0000, Richard Stobart wrote:
Dale,
If SHARED is true in a Service Object then the Service Object is not
re-entrant (because many clients are sharing its variables and therefore is
not multitasked (Funny logical naming if you ask me)). I got a lot of replies correcting me on what SHARED means! I do know
what SHARED means, I just phrased it backwards in my post. Excuse me
while I smack myself upside the head. :)
All replicates of a
load balanced partition are not re-entrant and thus equivalent to SHARED =
true. The advantage of load balancing is that the replicates can be
distributed over machines and thus the load is balanced. What I fail to understand is this: If you have a non-SHARED SO all by
itself in a partition which is not load-balanced, it will be
re-rentrant and multiple users can call it at the same time. But as
soon as you load-balance it, all of a sudden it behaves as if it were
SHARED. Why? I don't understand the technical limitations that impose
this, nor do I understand the advantage. For example, let's say that I
have 50 concurrent users of the SO I described above. It's in a
non-load-balanced partition, so all 50 users can access it at the same
time without any problem. Now let's say my server is a little
stressed, so I decide I want to load-balance my SO and have two
replicates, one on the original server and one on a second server. But
now that I've load-balanced it, the partitions act as if they're SHARED
and my 50 concurrent users are going to be lining up in queues and
suffering from horrible response times. How is this advantageous?
================================================
Dale V. Georg
Systems Analyst
Indus Consultancy Services
[email protected]
================================================

Dale,
The argument for load balancing services falls into two basic categories:
1. The service is blocking. That is, it is either not thread safe, not
signal tolerant, or not re-entrant. Most database libraries are the most
common example of this. Any partition that contains a non-multi-threaded
DB library will block the entire partition. Thus, you MUST load balance.
2. You want to spread the load across multiple servers. This is a tricky
one - you really have to do load testing to see if you're getting the
benefits you think you are.
Load balancing vs. multi-threading is a pretty hot topic among all of the
clients I visit. On a machine with native or certified POSIX threads and
multiple CPUs, you will certainly gain significant performance by not
load-balancing a non-blocking partition. Compare this with shared and/or
transactional services within a partition, non-native or non-POSIX thread
partitions, and machines with one CPU, and you have yourself quite a pot of
fish soup. Add another machine with differing properties, and the effects
can be exponential.
Then, there is also the notion that load balanced services are
intrinsically single threaded, which is not true. They can, and often do,
act as single threaded services because of the router. But it doesn't
imply that either the router, nor the partitions it routes, are necessarily
single threaded.
Don
At 07:38 PM 1/21/98, Dale V. Georg wrote:
On Wed, 21 Jan 1998 17:24:33 -0000, Richard Stobart wrote:
Dale,
If SHARED is true in a Service Object then the Service Object is not
re-entrant (because many clients are sharing its variables and therefore is
not multitasked (Funny logical naming if you ask me)). I got a lot of replies correcting me on what SHARED means! I do know
what SHARED means, I just phrased it backwards in my post. Excuse me
while I smack myself upside the head. :)
All replicates of a
load balanced partition are not re-entrant and thus equivalent to SHARED =
true. The advantage of load balancing is that the replicates can be
distributed over machines and thus the load is balanced. What I fail to understand is this: If you have a non-SHARED SO all by
itself in a partition which is not load-balanced, it will be
re-rentrant and multiple users can call it at the same time. But as
soon as you load-balance it, all of a sudden it behaves as if it were
SHARED. Why? I don't understand the technical limitations that impose
this, nor do I understand the advantage. For example, let's say that I
have 50 concurrent users of the SO I described above. It's in a
non-load-balanced partition, so all 50 users can access it at the same
time without any problem. Now let's say my server is a little
stressed, so I decide I want to load-balance my SO and have two
replicates, one on the original server and one on a second server. But
now that I've load-balanced it, the partitions act as if they're SHARED
and my 50 concurrent users are going to be lining up in queues and
suffering from horrible response times. How is this advantageous?
================================================
Dale V. Georg
Systems Analyst
Indus Consultancy Services
[email protected]
================================================
============================================
Don Nelson
Regional Consulting Manager - Rocky Mountain Region
Forte Software, Inc.
Denver, CO
Phone: 303-265-7709
Corporate voice mail: 510-986-3810
aka: [email protected]
============================================
"We tigers prefer to inflict excitement on others." - Hobbes

Similar Messages

  • Re: Shared SO's and Load Balancing

    Thanks, Tom. I understand this issue a lot better now.
    Dale
    On Wed, 21 Jan 1998 23:28:33 +0000, Tom O'Rourke wrote:
    Dale,
    You are right in that it many times does not make sense to load balance
    services that are multi-threaded.
    But, when a TOOL service object is partitioned with a DBSession service
    object, the TOOL service acts as if it is single-threaded regardless of if
    it is SHARED or not (while the partition is accessing the database). This
    means that the entire partition will be blocked (single-threaded) while the
    TOOL service object is using the DBSession to access a database.
    Forte blocks the partition to protect the integrity of the data being
    passed back and forth to the database because the database vendors have yet
    to provide a thread safe call interface to the RDBMS. This is changing as
    we speak and Forte is in the process of making the appropriate adjustments.
    So, this is the case where it makes complete sense to load balance a
    service that is not marked as being SHARED and why it can be a tremendous
    performance advantage to use Forte load balancing. As we all know, this
    architecture (TOOL EVSO partitioned with DBSession UVSO) is one that is
    widely used and proven to produce high performing applications.
    The point you bring up is a good one and that is often misunderstood. A
    load balanced SO will behave as if it is single threaded (or SHARED).
    FYI, we have just added a new feature in release 3.F. Performance-based
    load balancing. Check it out.
    Tom
    At 07:38 PM 1/21/98, Dale V. Georg wrote:
    On Wed, 21 Jan 1998 17:24:33 -0000, Richard Stobart wrote:
    Dale,
    If SHARED is true in a Service Object then the Service Object is not
    re-entrant (because many clients are sharing its variables and therefore is
    not multitasked (Funny logical naming if you ask me)). I got a lot of replies correcting me on what SHARED means! I do know
    what SHARED means, I just phrased it backwards in my post. Excuse me
    while I smack myself upside the head. :)
    All replicates of a
    load balanced partition are not re-entrant and thus equivalent to SHARED =
    true. The advantage of load balancing is that the replicates can be
    distributed over machines and thus the load is balanced. What I fail to understand is this: If you have a non-SHARED SO all by
    itself in a partition which is not load-balanced, it will be
    re-rentrant and multiple users can call it at the same time. But as
    soon as you load-balance it, all of a sudden it behaves as if it were
    SHARED. Why? I don't understand the technical limitations that impose
    this, nor do I understand the advantage. For example, let's say that I
    have 50 concurrent users of the SO I described above. It's in a
    non-load-balanced partition, so all 50 users can access it at the same
    time without any problem. Now let's say my server is a little
    stressed, so I decide I want to load-balance my SO and have two
    replicates, one on the original server and one on a second server. But
    now that I've load-balanced it, the partitions act as if they're SHARED
    and my 50 concurrent users are going to be lining up in queues and
    suffering from horrible response times. How is this advantageous?
    ================================================
    Dale V. Georg
    Systems Analyst
    Indus Consultancy Services
    [email protected]
    ================================================
    ================================================
    Dale V. Georg
    Systems Analyst
    Indus Consultancy Services
    [email protected]
    ================================================

    Thanks, Tom. I understand this issue a lot better now.
    Dale
    On Wed, 21 Jan 1998 23:28:33 +0000, Tom O'Rourke wrote:
    Dale,
    You are right in that it many times does not make sense to load balance
    services that are multi-threaded.
    But, when a TOOL service object is partitioned with a DBSession service
    object, the TOOL service acts as if it is single-threaded regardless of if
    it is SHARED or not (while the partition is accessing the database). This
    means that the entire partition will be blocked (single-threaded) while the
    TOOL service object is using the DBSession to access a database.
    Forte blocks the partition to protect the integrity of the data being
    passed back and forth to the database because the database vendors have yet
    to provide a thread safe call interface to the RDBMS. This is changing as
    we speak and Forte is in the process of making the appropriate adjustments.
    So, this is the case where it makes complete sense to load balance a
    service that is not marked as being SHARED and why it can be a tremendous
    performance advantage to use Forte load balancing. As we all know, this
    architecture (TOOL EVSO partitioned with DBSession UVSO) is one that is
    widely used and proven to produce high performing applications.
    The point you bring up is a good one and that is often misunderstood. A
    load balanced SO will behave as if it is single threaded (or SHARED).
    FYI, we have just added a new feature in release 3.F. Performance-based
    load balancing. Check it out.
    Tom
    At 07:38 PM 1/21/98, Dale V. Georg wrote:
    On Wed, 21 Jan 1998 17:24:33 -0000, Richard Stobart wrote:
    Dale,
    If SHARED is true in a Service Object then the Service Object is not
    re-entrant (because many clients are sharing its variables and therefore is
    not multitasked (Funny logical naming if you ask me)). I got a lot of replies correcting me on what SHARED means! I do know
    what SHARED means, I just phrased it backwards in my post. Excuse me
    while I smack myself upside the head. :)
    All replicates of a
    load balanced partition are not re-entrant and thus equivalent to SHARED =
    true. The advantage of load balancing is that the replicates can be
    distributed over machines and thus the load is balanced. What I fail to understand is this: If you have a non-SHARED SO all by
    itself in a partition which is not load-balanced, it will be
    re-rentrant and multiple users can call it at the same time. But as
    soon as you load-balance it, all of a sudden it behaves as if it were
    SHARED. Why? I don't understand the technical limitations that impose
    this, nor do I understand the advantage. For example, let's say that I
    have 50 concurrent users of the SO I described above. It's in a
    non-load-balanced partition, so all 50 users can access it at the same
    time without any problem. Now let's say my server is a little
    stressed, so I decide I want to load-balance my SO and have two
    replicates, one on the original server and one on a second server. But
    now that I've load-balanced it, the partitions act as if they're SHARED
    and my 50 concurrent users are going to be lining up in queues and
    suffering from horrible response times. How is this advantageous?
    ================================================
    Dale V. Georg
    Systems Analyst
    Indus Consultancy Services
    [email protected]
    ================================================
    ================================================
    Dale V. Georg
    Systems Analyst
    Indus Consultancy Services
    [email protected]
    ================================================

  • Reverse Proxy and Load Balancer for SMP 2.3 and Agentry Application

    Hi Expert,
    I'm putting in place a mobile solution composed by SMP 2.3 SPS 4 and SAP ECC 6.0. In the SMP 2.3 I created the agentry server and I have deployed my agentry application.
    My SMP/Agentry infrastructure is composed by two servers therefore I need a load balancer for balance the load into the several servers. Furthermore I need to use a reverse proxy in my DMZ zone.
    Based on what indicated in the SAP note "1904213 - SAP Mobile Platform Server Release Information" the Apache Reverse Proxy is not supported for Agentry clients. Agentry uses nginx for Reverse Proxy.
    I also found the following document How-to-Guide for Reverse Proxy and Load Balancing in SAP Mobile Platform 3.x that explain how to set-up a reverse proxy and load balancer with nginx and apache.
    Both the SAP note and the HOW to document are refereed to SMP 3.0 and not to SMP 2.3.
    I would know if the NGINX must be used also for SMP 2.3.
    Any suggestion/information is appreciated.
    Thanks in advance
    g.

    Please see Agentry Network Landscapes

  • Cache and Load Balancing with Oracle APEX Listener

    Hi,
    I intend to use only HTTP access.
    How to implement a Cache and Load Balancing with the Oracle APEX Listener?
    Is it possible to do with the the standalone running APEX Listener?
    Thanks by advance for any tips/documentation/references.
    Kind Regards.

    Hi,
    I think this question is best asked in the APEX Listener forum:
    ORDS, SODA & JSON in the Database
    Kind regards
    Sandro

  • Cache and Load Balancing for the Oracle APEX Listener

    Hi,
    I intend to use only HTTP access.
    My database is Oracle 11gR2, SE, 32 bit.
    How to implement a Cache and Load Balancing with the Oracle APEX Listener?
    Is it possible to do with the the standalone running APEX Listener?
    Thanks by advance for any tips/documentation/references.
    Kind Regards.

    Error. To be closed.

  • Cache and Load Balancing for Oracle APEX Listener

    Hi,
    I intend to use only HTTP access.
    The database I use is Oracle11gR2 SE 32bit.
    How to implement a Cache and Load Balancing with the Oracle APEX Listener?
    Is it possible to do with the the standalone running APEX Listener?
    Thanks by advance for any tips/documentation/references.
    Kind Regards.

    Error. To be closed.

  • PIX Redundant Internet Line and Load balancing

    I would like to find out if it's possible to configure my Cisco PIX 525 to use a secondary internet line from a different provider and perform load balancing. I'm using PIX Version 6.3(1)

    PIX version 6.3 does not support Redundancy and load balancing. but PIX/ ASA with version 7.0 supports Redundancy.

  • VPN device with dual ISP, fail-over, and load balancing

    We currently service a client that has a PIX firewall that connects to multiple, separate outside vendors via IPSEC VPN. The VPN connections are mission critical and if for any reason the VPN device or the internet connection (currently only a T1) goes down, the business goes down too. We're looking for a solution that allows dual-ISP, failover, and load balancing. I see that there are several ASA models as well as the IOS that support this but what I'm confused about is what are the requirements for the other end of the VPN, keeping in mind that the other end will always be an outside vendor and out of our control. Current VPN endpoints for outside vendors are to devices like VPN 3000 Concentrator, Sonicwall, etc. that likely do not support any type of fail-over, trunking, load-balancing. Is this just not possible?

    Unless I am mistaken the ASA doesn't do VPN Load Balancing for point-to-point IPSec connections either. What you're really after is opportunistic connection failover, and/or something like DMVPN. Coordinating opportunistic failover shouldn't be too much of an issue with the partners, but be prepared for lot of questions.

  • Data Centre Interconnection - firewall and load balancer deployment

    Hi all,
    I've read lots of Cisco docs/white papers on DCI - Layer 2 extension between DCs, but as yet I cannot find any decent information on how best to deploy firewalls and load balancers in such a design. I've seen refs to FHRP isolation on Nexus 7k (and possible 6k if you use DCI block) but nothing on the services elements.
    The services element seems to be a complete minefield here:
    - active/standby across sites, or deploy resilient pairs in each site?
    - how to align optimal traffic flows inbound and ooutbound (RHI, SNAT, etc.)
    - best practice suggestions ideally.
    Cisco DCI docs seem to always gloss over the fact that most customers would have to deal with firewalls and load balancers here, and simply refer to 'coming soon' for that info.
    If anyone has any good suggestions/links to docs explaining detailed implementation info would be much appreciate
    Thanks
    Phil

    You might want to check out this new product called ITD.
    Simple and faster solution:
    ITD provides :
    ASIC based multi-terabit/s L3/L4 load-balancing at line-rate
    No service module or external L3/L4 load-balancer needed. Every N7k port can be used as load-balancer.
    Redirect line-rate traffic to any devices, for example web cache engines, Web Accelerator Engines (WAE), video-caches, etc.
    Capability to create clusters of devices, for example, Firewalls, Intrusion Prevention System (IPS), or Web Application Firewall (WAF), Hadoop cluster
    IP-stickiness
    Resilient (like resilient ECMP)
    VIP based L4 load-balancing
    NAT (available for EFT/PoC). Allows non-DSR deployments.
    Weighted load-balancing
    Load-balances to large number of devices/servers
    ACL along with redirection and load balancing simultaneously.
    Bi-directional flow-coherency. Traffic from A-->B and B-->A goes to same node.
    Order of magnitude OPEX savings : reduction in configuration, and ease of deployment
    Order of magnitude CAPEX savings : Wiring, Power, Rackspace and Cost savings
    The servers/appliances don’t have to be directly connected to N7k
    Monitoring the health of servers/appliances.
    N + M redundancy.
    Automatic failure handling of servers/appliances.
    VRF support, vPC support, VDC support
    Supported on both Nexus 7000 and Nexus 7700 series.
    Supports both IPv4 and IPv6
    N5k / N6k support : coming soon
    Blog
    At a glance
    ITD config guide
    Email Query or feedback:[email protected]

  • What does per Wlan Band select and load balancing do ?

    Good morning.....We recently upgraded our controllers from 4.2.185 to 6.0.188 and have noticed many clients having connectivity issues. We have Aggressive load balancing turned off globally but have noticed that band select and load balancing are enabled on the
    Wlan. Are these settings mutually exclusive or do they do the same thing ?  Does the Wlan setting override the default ? We have noticed that there is
    output doing "debug dot11 load-balancing"
    Thanx.....Dave

    I believe we never had load balancing turned on when running 5.2 code. We jumped from 5.2 to 6.x temporarily and then to 7.0 within a 30 day time frame this summer. We're a large university and we had very few users on WiFi during that time.
    The Macintosh laptops are having nothing but trouble since school began, and I have gone over everything and found that band select is turned on as well as load balancing. Since band select didn't exist in 5.2 (I believe) I know it wasn't on. As for load balancing, I don't believe it was on, and I discovered it was turned on when recently reviewing our configs.
    The Macintosh laptops have been debugged and our Mac gurus tell us they're getting a message that equates to "the AP is busy, or the AP is full". This leads me to believe that load balancing got turned on during the upgrade and we didn't notice, which caused the Macintoshes to have issues.
    We don't have any VoWiFi clients so we don't have to support them, and we don't officially support smartphones, either.
    I turned off load balancing and will see how it goes....
    Thanks!

  • ARFC: Single Server and Load Balancing

    Hi All,
    I am trying to create aRFC model. In SAP logon screen, I can see two tab pages - Single Server and Load Balancing.
    Can you please let me know when we have to use which tab?
    Thanks
    TG

    Single Server Connect or Load Balancing connect is completely independend from the location where SAP Gui Client is installed.
    Single Server connect means that your are directly connecting to an ABAP Server using hostname and systemnumber you have to provide.
    Load Balancing Connect means that you specify the message server of the central instance of an ABAP Server group. The SAPGUI first connects to the message server which will provide the SAPGUI with the information about the best performing ABAP server. SAPGUI will then connect to this ABAP server.
    Single Server is suitable for small landscapes with lets say less than 4 application servers. In huger configurations (and those which I know will grow to more than 3 servers)I would prefer to use logon groups - aka Load Balancing.
    Peter

  • Reverse Proxy plug in and Load Balancer Plug in

    Hi,
    Can anyone please provide me with an example obj.conf file showing how to combine the reverse proxy plug-in and Load Balancer plug-in.
    I would like to use the reverse proxy plug in to detect when static content is requested and provide this from the web server. Requests for dynamic content would then be forwarded to an Application server via the Load balancer plug-in. I have found plenty of documentation on how to configure these plug-in separately but nothing on how to combine the two.

    smiking
    reverse proxy plugin - its job is to forward the requests to another server for a specific task. you can use the webserver 7 . it does forward and limited load balancing (using round robin ) based on the number of servers you provide in the configuration. i would say this is a poor man's setup.
    load balancer plugin - some app servers like sun java system app server or web logic provide this plugin so that you can effectively use the back end app server
    with both these setup, you can <if> constructs to determine which requests need to be forwarded to the back end server.
    I wonder, why do you need both - if both of them is designed to do the same thing.

  • Web Dispatcher - Reverse Proxy and Load Balancing

    I'm finding limited docs on Web Dispatcher with regard to reverse proxy and load balancing.  Are you aware of some recent presentations or docs in this area?  The info on help.sap.com is not what I'm looking for.
    Thanks.

    Hi,
    best thing is that you look at your scenarios and test the web dispatcher against each of it, like:
    - SSL
    - Portal only
    - Web Dynpro ABAP / Java
    - BSP
    - Different backend systems like SRM, MDM
    - Several backends with 1 Web Dispatcher
    After getting a list of use cases that you can test quite easily (installation of Web Dispatcher is done fast and can be done on a local PC), you can contact SAP Support and ask them about the specific problems and questions you encountered. This way, you'll get the official answer, sometimes they will even inform you about "secret" parameters and options.
    As of the reverse proxy functionality: there are several version of Web Dispatcher available that differ from the functionality offered. The latest version - 7.2 - is the one that offers the most, i.e. allows you to create rewrite rules like Apache.
    SAP Note 908097 - SAP Web Dispatcher: Released releases and applying patches
    br,
    Tobias

  • What SSL accelerator and load-balancer does anyone recommend?

    Hi:
    I wanted to find out:
    Does anyone recommend SSL accelerator cards/boards or SSL accelerator appliances?
    What SSL accelerator and load balancer does aynone recommend to help 9iAS?

    Ana_Alm wrote:
    Hi there!
    I just downloaded and installed OS X Lion, and I'm loving it so far.
    However, I've seen that Mountain Lion will have some new features when it comes to social apps (what I call the ones that combine twitter, facebook, rss readers and so on).
    So, does anyone knows any cools apps for that? I'm currently using Socialite, that combines all those three, but it has a few issues I don't particularly like. Plus, I'm using Adium for a msn client. I'm also thinking about downloading that beta version of "Messages" that will be realeased on Mountain Lion.
    So, what do you think? Give me your ideas
    Thanks a lot in advance!
    As Mountain Lion has not been released to the public yet, then most of us have no idea which companies have updated the development of their Apps for  ML. It is in Development phase so any App you try is at your own risk.
    Good Luck
    Pete

  • Redundant and load balance of Final Cut Server

    Greeting All,
    Anyone has good start of redundant and load balance solution for Final Cut Server? Please advise!!
    Thanks,
    Spin

    Found this and seems to have worked...
    http://support.apple.com/kb/HT3836
    Just ran those two commands and then restarted the web service and now I get the page.

Maybe you are looking for