[cisco-voip] How Important is Running "Supported?"

[Anthony Holloway]

Well said Mike, I like your point of view.

I have to be honest, I didn't know that about IE9.  Not that it matters for
my particular case, because we're upgrading to IE11 now anyways, but, it's
good to know that no one should be on IE9 for any reason today.

I take this opportunity to say too, that IE9 is only one of several
browsers that could have been used with Finesse, and caused the CPU issue.
See the Browser implementation
<https://en.wikipedia.org/wiki/WebSocket#Browser_implementation> section of
the WebSockets wikipage for a list of browser which support this feature.
So, in theory, the client could have been on a Cisco unsupported browser
other than IE9, and still seen the same issue.

Actually, I take that back, it looks like almost all major browsers support
WebSockets <http://caniuse.com/#search=websocket>.  That doesn't leave a
lot of room for choice of browsers that would fail a WebSockets test.
There goes that argument.

Thanks again for your reply.

On Thu, May 12, 2016 at 4:45 PM, Norton, Mike <mikenorton at pwsd76.ab.ca>
wrote:

> When deciding whether to venture into unsupported territory, I think it is
> worth distinguishing between different types of “unsupported.”
>
>
>
> One type of “unsupported” is when a vendor says certain pieces must be
> certain things and at certain versions, but you choose to go against their
> requirement.
>
>
>
> Another type of “unsupported” is when a vendor declares that they are
> discontinuing all maintenance/development on a certain piece and that it
> must be migrated away from immediately.
>
>
>
> Your situation is both. Cisco says browser must be a certain
> flavour/version to work with their particular solution, *and* Microsoft
> told EVERYBODY to get off IE9 months ago.
>
>
>
> There are lots of environments where intentionally going “unsupported” is
> an acceptable risk, but IMO mixing both types of unsupported simultaneously
> is probably asking for trouble in almost all cases.
>
>
>
> -mn
>
>
>
>
>
> *From:* cisco-voip [mailto:cisco-voip-bounces at puck.nether.net] *On Behalf
> Of *Anthony Holloway
> *Sent:* May-12-16 10:07 AM
> *To:* Cisco VoIP Group <cisco-voip at puck.nether.net>
> *Subject:* [cisco-voip] How Important is Running "Supported?"
>
>
>
> All,
>
>
>
> Over the past 10 years, I have seen a multitude of deployments that were
> not in a 100% supported configuration.  Most of the time, this simply
> results in "don't ask, don't tell" and as long as you didn't need TAC
> support, everything should be fine.  I don't necessarily mean UCCX server
> compatibility with CUCM, but like phone models, firmware, IOS code, gateway
> models, web browsers, OS, Agent shared lines or LG membership, etc.
>
>
>
> Well, recently I just ran into a non-supported setup, where it caused the
> server to hit 100% CPU utilization and cause all sorts of problems in the
> application.
>
>
>
> It was UCCX v11 and using IE9 for Finesse.
>
>
>
> You can read more about the difference in the browser here, if you have
> access:
>
> https://communities.cisco.com/message/215058/
>
>
>
> Otherwise the summary is basically this:
>
> UCCX v11 introduced a new technology to support Live Data feeds from
> Finesse to CUIC: Socket.IOhttp://socket.io/ <http://Socket.IO>.
> Socket.IO uses HTTP WebSockets <https://en.wikipedia.org/wiki/WebSocket>
> to open a single TCP connection from Finesse to UCCX, and then passes all
> updates to the data via this single connection.
>
>
>
> IE9 lacks the feature to support WebSockets and the Socket.IO software
> automatically allows older browser clients to fallback to single HTTP GET
> <http://stackoverflow.com/questions/12993704/ie-and-socket-io-compatibility>
> calls for each data refresh interval.
>
>
>
> This means that instead of having 100 Agents using 1 socket each, for a
> total of 100 connections that are persistent, you'll end up 100 Agents
> using hundreds of connections (quickly setup and then closed) throughout
> the day, resulting in what is essentially a DoS attack on the SocketIO
> service.
>
>
>
> In theory, this would happen with any browser that doesn't support
> WebSockets.
>
>
>
> So, how important is being in compliance when it comes to what's supported
> and not supported?  Do you typically bend the rules, or are you rigid and
> strict?
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20160513/d5e69a71/attachment.html>