[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Highly available web applications



----- Original Message -----
> From: "meghdoot bhattacharya" <meghdoot_b yahoo com>
> To: dev lists openshift redhat com
> Sent: Friday, July 5, 2013 3:02:19 AM
> Subject: Re: Highly available web applications
> 
> The PEP mentions this HA model only for web applications when the same model
> should apply also for web services residing in traditional mid-tier or
> should it not?
> What about non HTTP based apps, since the routing layer can directly reach
> out to Haproxy that can be configured in TCP LB mode, can the same model
> apply for primary non HTTP based carts?
> 

Non http requests don't support the Host Header (for the most part) so we left them out of scope for that pep.  Additionally backend services like databases, etc, will be made highly available in their own tier but the web request / routing tier won't be accessing the data directly.

As far as setting the front end load balancers to TCP mode.  You could do that but you'd have to setup a new load balancer for every new service you add except for a few specific use cases.

    -Mike

> Thx
> 
> 
> ________________________________
>  From: "dev-request lists openshift redhat com"
>  <dev-request lists openshift redhat com>
> To: dev lists openshift redhat com
> Sent: Thursday, July 4, 2013 5:52 PM
> Subject: dev Digest, Vol 16, Issue 8
>  
> 
> Send dev mailing list submissions to
>     dev lists openshift redhat com
> 
> To subscribe or unsubscribe via the World Wide Web, visit
>     http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> or, via email, send a message with subject or body 'help' to
>     dev-request lists openshift redhat com
> 
> You can reach the person managing the list at
>     dev-owner lists openshift redhat com
> 
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of dev digest..."
> 
> 
> Today's Topics:
> 
>    1. Re: Draft 1 of PEP005 - Highly available web applications -
>       is    available for review (Thomas Wiest)
>    2. Re: Draft 1 of PEP005 - Highly available web applications -
>       is    available for review (Clayton Coleman)
>    3. REST service messages: is severity unique? (Andr? Dietisheim)
>    4. Re: REST service messages: is severity unique? (Clayton Coleman)
>    5. Re: REST service messages: is severity unique? (Daniel McPherson)
>    6. Re: REST service messages: is severity unique? (Clayton Coleman)
> 
> 
> ----------------------------------------------------------------------
> 
> Message: 1
> Date: Thu, 04 Jul 2013 13:15:50 -0400
> From: Thomas Wiest <twiest redhat com>
> To: Clayton Coleman <ccoleman redhat com>
> Cc: "dev lists openshift redhat com" <dev lists openshift redhat com>
> Subject: Re: Draft 1 of PEP005 - Highly available web applications -
>     is    available for review
> Message-ID: <51D5ADC6 8030908 redhat com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> Very interesting PEP.
> 
> In the PEP, you say that the proxy layer would likely be sharded in some
> fashion for large deploys.
> 
> Is the sharding something that the broker will manage?
> 
> Thomas
> 
> 
> On 07/03/2013 07:59 PM, Clayton Coleman wrote:
> > As part of the long term evolution of the OpenShift platform, we intend to
> > introduce support for highly available web applications.  A PEP has been
> > drafted in the public repository to describe some of the design
> > considerations and general intent:
> >
> >    https://github.com/openshift/openshift-pep/blob/master/openshift-pep-005.md
> >
> > This is an early draft, and there are topics that are deliberately
> > considered out of scope of the document at the current time, or are
> > insufficiently specified.  If you have questions, comments, or feedback,
> > please reply to this thread.
> >
> > Thanks!
> >
> > _______________________________________________
> > dev mailing list
> > dev lists openshift redhat com
> > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> 
> 
> 
> ------------------------------
> 
> Message: 2
> Date: Thu, 4 Jul 2013 13:28:49 -0400 (EDT)
> From: Clayton Coleman <ccoleman redhat com>
> To: Thomas Wiest <twiest redhat com>
> Cc: "dev lists openshift redhat com" <dev lists openshift redhat com>
> Subject: Re: Draft 1 of PEP005 - Highly available web applications -
>     is    available for review
> Message-ID: <21E69A0B-5C37-49DE-A035-C8AED139EEDA redhat com>
> Content-Type: text/plain;    charset=us-ascii
> 
> Depends, but I would say it might be better to put that intelligence in the
> component that handles app create events.  However, there's nothing
> preventing the plugin receiving the routing spi events from making a
> decision and recording it in the brokers mongo.
> 
> Arguably you need to make the sharding decision based on the current volume
> of traffic flowing through that router.  For instance, a router that is
> servicing a high volume app (50+ gears, 1k req/sec) might not be the best
> place to colocate another high volume app - it may be an operational
> responsibility to periodically rebalance proxies manually or via scripts.
> 
> The assumption is that the routers must scale to handle the highest possible
> individual volume app deployed in Openshift - well tuned proxies should be
> able to handle hundreds of thousands of requests a second, but the placement
> of a high volume app is going to be an important decision.
> 
> This part of the pep is a bit vague - mike was planning on flushing out the
> details of that more.
> 
> On Jul 4, 2013, at 1:15 PM, Thomas Wiest <twiest redhat com> wrote:
> 
> > Very interesting PEP.
> > 
> > In the PEP, you say that the proxy layer would likely be sharded in some
> > fashion for large deploys.
> > 
> > Is the sharding something that the broker will manage?
> > 
> > Thomas
> > 
> > 
> > On 07/03/2013 07:59 PM, Clayton Coleman wrote:
> >> As part of the long term evolution of the OpenShift platform, we intend to
> >> introduce support for highly available web applications.  A PEP has been
> >> drafted in the public repository to describe some of the design
> >> considerations and general intent:
> >> 
> >>  https://github.com/openshift/openshift-pep/blob/master/openshift-pep-005.md
> >> 
> >> This is an early draft, and there are topics that are deliberately
> >> considered out of scope of the document at the current time, or are
> >> insufficiently specified.  If you have questions, comments, or feedback,
> >> please reply to this thread.
> >> 
> >> Thanks!
> >> 
> >> _______________________________________________
> >> dev mailing list
> >> dev lists openshift redhat com
> >> http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> > 
> 
> 
> 
> ------------------------------
> 
> Message: 3
> Date: Fri, 05 Jul 2013 01:04:04 +0200
> From: Andr? Dietisheim <adietish redhat com>
> To: "dev lists openshift redhat com" <dev lists openshift redhat com>,
>     Daniel McPherson <dmcphers redhat com>
> Subject: REST service messages: is severity unique?
> Message-ID: <51D5FF64 5020500 redhat com>
> Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"
> 
> Hi
> 
> I'm currently trying to have the Eclipse tooling for OpenShift consuming
> the old and new json when it comes to messages (no more "result" field,
> "result" now present in severity -
> https://bugzilla.redhat.com/show_bug.cgi?id=980928). We badly hope for a
> hotfix since we're GA in about 1 week.
> In the meantime, I'm trying to consume the new and old way at the same
> time (we need to be able to talk to OSE, too, cannot simply stick to the
> new format).
> I was thus wondering if severity was unique among messages for the same
> field or is it possible to get several messages with the same field and
> severity?
> ex. there are several messages with field "undefined", are the also
> eventually several messages with severity "result"?
> 
> Thanks for your pointers and hints!
> Cheers
> Andr?
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> <http://lists.openshift.redhat.com/openshift-archives/dev/attachments/20130705/f0cce7ac/attachment.html>
> 
> ------------------------------
> 
> Message: 4
> Date: Thu, 4 Jul 2013 19:26:43 -0400 (EDT)
> From: Clayton Coleman <ccoleman redhat com>
> To: Andr? Dietisheim <adietish redhat com>
> Cc: "dev lists openshift redhat com" <dev lists openshift redhat com>
> Subject: Re: REST service messages: is severity unique?
> Message-ID: <3E73D2C7-9798-4267-92B7-6A2E9E142BCB redhat com>
> Content-Type: text/plain; charset="utf-8"
> 
> Is the field value a string "undefined" or just null?
> 
> On Jul 4, 2013, at 7:04 PM, Andr? Dietisheim <adietish redhat com> wrote:
> 
> > Hi
> > 
> > I'm currently trying to have the Eclipse tooling for OpenShift consuming
> > the old and new json when it comes to messages (no more "result" field,
> > "result" now present in severity -
> >   https://bugzilla.redhat.com/show_bug.cgi?id=980928). We badly hope for a
> > hotfix since we're GA in about 1 week.
> > In the meantime, I'm trying to consume the new and old way at the same time
> > (we need to be able to talk to OSE, too, cannot simply stick to the new
> > format).
> > I was thus wondering if severity was unique among messages for the same
> > field or is it possible to get several messages with the same field and
> > severity?
> > ex. there are several messages with field "undefined", are the also
> > eventually several messages with severity "result"?
> > 
> > Thanks for your pointers and hints!
> > Cheers
> > Andr?
> > _______________________________________________
> > dev mailing list
> > dev lists openshift redhat com
> > http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> <http://lists.openshift.redhat.com/openshift-archives/dev/attachments/20130704/8445bb80/attachment.html>
> 
> ------------------------------
> 
> Message: 5
> Date: Thu, 4 Jul 2013 20:20:17 -0400 (EDT)
> From: Daniel McPherson <dmcphers redhat com>
> To: Clayton Coleman <ccoleman redhat com>
> Cc: "dev lists openshift redhat com" <dev lists openshift redhat com>
> Subject: Re: REST service messages: is severity unique?
> Message-ID: <9A5BDDDB-830D-4054-87CD-F16613D1E5DD redhat com>
> Content-Type: text/plain; charset="utf-8"
> 
> Adding Abhishek and Lili.
> 
> On Jul 4, 2013, at 7:26 PM, Clayton Coleman <ccoleman redhat com> wrote:
> 
> > Is the field value a string "undefined" or just null?
> > 
> > On Jul 4, 2013, at 7:04 PM, Andr? Dietisheim <adietish redhat com> wrote:
> > 
> >> Hi
> >> 
> >> I'm currently trying to have the Eclipse tooling for OpenShift consuming
> >> the old and new json when it comes to messages (no more "result" field,
> >> "result" now present in severity -
> >>   https://bugzilla.redhat.com/show_bug.cgi?id=980928). We badly hope for a
> >> hotfix since we're GA in about 1 week.
> >> In the meantime, I'm trying to consume the new and old way at the     same
> >> time (we need to be able to talk to OSE, too, cannot simply stick to the
> >> new format).
> >> I was thus wondering if severity was unique among messages for the same
> >> field or is it possible to get several messages with the same field and
> >> severity?
> >> ex. there are several messages with field "undefined", are the also
> >> eventually several messages with severity "result"?
> >> 
> >> Thanks for your pointers and hints!
> >> Cheers
> >> Andr?
> >> _______________________________________________
> >> dev mailing list
> >> dev lists openshift redhat com
> >> http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> <http://lists.openshift.redhat.com/openshift-archives/dev/attachments/20130704/a297db14/attachment.html>
> 
> ------------------------------
> 
> Message: 6
> Date: Thu, 4 Jul 2013 20:52:31 -0400 (EDT)
> From: Clayton Coleman <ccoleman redhat com>
> To: Daniel McPherson <dmcphers redhat com>
> Cc: "dev lists openshift redhat com" <dev lists openshift redhat com>
> Subject: Re: REST service messages: is severity unique?
> Message-ID: <0F90166C-15A3-4884-BE47-A1FCBA4F7948 redhat com>
> Content-Type: text/plain; charset="utf-8"
> 
> It's definitely a regression to public API behavior.  We worked around it in
> rhc but it would be broken for older versions.
> 
> On Jul 4, 2013, at 8:20 PM, Daniel McPherson <dmcphers redhat com> wrote:
> 
> > Adding Abhishek and Lili.
> > 
> > On Jul 4, 2013, at 7:26 PM, Clayton Coleman <ccoleman redhat com> wrote:
> > 
> >> Is the field value a string "undefined" or just null?
> >> 
> >> On Jul 4, 2013, at 7:04 PM, Andr? Dietisheim <adietish redhat com> wrote:
> >> 
> >>> Hi
> >>> 
> >>> I'm currently trying to have the Eclipse tooling for OpenShift consuming
> >>> the old and new json when it comes to messages (no more "result" field,
> >>> "result" now present in severity -
> >>>   https://bugzilla.redhat.com/show_bug.cgi?id=980928). We badly hope for
> >>> a hotfix since we're GA in about 1 week.
> >>> In the meantime, I'm trying to consume the new and old way at the same
> >>> time (we need to be able to talk to OSE, too, cannot simply stick to the
> >>> new format).
> >>> I was thus wondering if severity was unique among messages for the same
> >>> field or is it possible to get several messages with the same field and
> >>> severity?
> >>> ex. there are several messages with field "undefined", are the also
> >>> eventually several messages with severity "result"?
> >>> 
> >>> Thanks for your pointers and hints!
> >>> Cheers
> >>> Andr?
> >>> _______________________________________________
> >>> dev mailing list
> >>> dev lists openshift redhat com
> >>> http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL:
> <http://lists.openshift.redhat.com/openshift-archives/dev/attachments/20130704/fb910920/attachment.html>
> 
> ------------------------------
> 
> _______________________________________________
> dev mailing list
> dev lists openshift redhat com
> http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> 
> 
> End of dev Digest, Vol 16, Issue 8
> **********************************
> _______________________________________________
> dev mailing list
> dev lists openshift redhat com
> http://lists.openshift.redhat.com/openshiftmm/listinfo/dev
> 


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]