carl

For voice, 150 ms is generally accepted as the largest roundtrip latency that is acceptable for two-way conversation. It's a real problem when doing VoIP.

[Edited by carl - 4/1/2003 9:50:52 AM]

super_si

Times come for another Assignment ive given you lot 4months off

Potential response times for specific network applications

Networked word processor
On-line transaction processing system run from a terminal on a centralised system
Email
External database (located on another continent)

Network reliability performance standards

Component Required standard System requirements
WAN communication lines WAN lines required to be available 100% of time Multiple leased lines or sharing arrangements with other companies in the event of a line failure
Hard disk drives Network servers to be available 100% of time
Network file server File server capable of being replaced within 5 minutes of failure

ive searched the damm web , but i cant find anything of any relevance on this.

Theres system recovery too but it'll take ages to post it up
Here Section 3

Thanks once again

Si



[Edited by super_si - 3/31/2003 4:05:01 PM]

SiCotty

Si, Whoever gave you the assignment knows bugger all about WAN links. No compnay will ever give you a 100% SLA. You are looking at potentially 99.99 or 99.999 but never 100%. Normally most clients settle with a WAN link and ISDN backup (if the apps can take it). If you bring in 2x leased lines then you have to be sure that they don't come into the building the same way and are connected to two different POPS. This is very expensive.

If the application is a server based application then latency is the key. You need to have some sort of queuing on the WAN router which lets small packets go before large packets so a large FTP session will not swamp a telnet session for example. You might want to have a look at Citrix.

For the servers something like a cluster attached to a high availability disk array. Cluster has multiple connections to the network using either load balancing on the cluster or within the network. You could connect all your servers to a SAN etc.

You might get 100% uptime SLA with a clause in the contract saying that any sheduled downtime or disruptions of under 10 minutes does not count

Simon

[Edited by SiCotty - 3/31/2003 4:58:39 PM]

super_si

Cheers mate, hes a ****ing w@nker thats who.

ive should another mate off here, an assignment. He pissed him self and laughed his **** off at some the bull**** they come out with.

Thanks anyway, need to fill some **** in to keep him happy though

Si

carl

Should be possible to give a 100% SLA under a single failure condition. As well as multiple redundant links, you'll need to ensure that the links don't share the same cable duct, each comes into the building separately and they use separate risers.

super_si

What about the other questions

ChrisB

Instead of queing on the WAN router, something like a Packeteer traffic shaper would help ensure application traffic is prioritised over FTP/HTTP etc.

We had redundant SDH fibre feeds into our old office for 'Net traffic. IIRC, BT rules said the cables could never be closer than 1 metre and that was a point of delivery. As previously mentioned, they had to route to seperate POPs.

In addition to the lines, you'll need redundaunt routers (so something like Cisco HSRP) and power supplies with backups.

dsmith

Actually I know of one (unfortuneately) that does. It comes down not to what is technically acheivable but what is considered an acceptable business risk for the end-to-end service.

i.e. Cost of providing network with genuine 100% up-time vs Cost of 99.99% Network + Service Credits for the other 0.01% of the time.

Deano

super_si

This blokes not even and IT teacher hes a fking business tw@t.

Si

dsmith

There you go - point proved

ferkin business tw@ts fundamentally dont understand IT. I think your course seems like an excellent introduction to real world IT Si

Deano

SiCotty

Deano,

I forgot about that way of working an SLA, I suppose it all depends upon the amount of business you get from that customer and over what time period you have locked them into.

Si I would include this as he would understand the business behind offering such an SLA. You could also make some assumptions about the amount of business your company would loose if it went for a less tight SLA and suffered an outage. Then compared the two.

Come to think of it I know of a Company that will insure your service against possible downtime if you host it with them. So then you have an argument of outsourcing your server hosting requirements. Use the argument about setting up your own support staff and infrastructure as opposed to buying the service from somebody with a custom built facility and 24x7 service and support.

Giving it a business slant as well as a technical slant might get you extra marks.

Simon

[Edited by SiCotty - 3/31/2003 5:40:09 PM]

RoadrunnerV2

Your teacher must be a right tw@t How can you calculate the answers with only the following info

Networked word processor 10 secs
On-line transaction processing system run from a terminal on a centralised system 8.7645 secs
Email 1 sec unless **** attachment 10mins
External database (located on another continent) hard one this, reckon 15mins should cover that

Unbelievable

[Edited by RoadrunnerV2 - 3/31/2003 5:40:04 PM]

super_si

Whats SAL lol.

Thats what i thought.I quote

i cant open ms word in that lol

SiCotty

Just had a read of the assignment. The main point is user perception of the service uptime. If they work from 9-5 and the service is always available then from their perception the service is 100% reliable. If you have downtime between those times then who is to know.

The same is true regarding response times, with faster computers and peoples need for information as soon as they press a button the perception is always they need an answer as soon as they hit the button. This contrasts to earlier systems when uses waited for seconds to get responses back from a system but still though it was fast.

With regards to the last bit on voice/video quality etc I will have a look to see if I can dig anything out but you might want to look at voice and video codecs. Examples for voice are G729, G711 and G723. You also have something called a mean opinion score or MOP to do with Toll quaility voice. You could arguee that people percenption of what is a good quality phone call has been significantly lowered by the use of Mobile Phones.

For video you might want to check out microsoft, apple and realaudio web sites for examples.

Si

SiCotty

Just noticed the submission date, What have you been up to for the last few months

Si

super_si

to55ing it off what ever else lol

Si

super_si

any ideas on Network recoverability performance standards