From: hcb@gettcomm.com
Date: Fri Dec 19 2003 - 23:18:16 GMT-3
Quoting "McNeace, Roger" <RMcNeace@ciena.com>:
> Five 9's SLA equates to 5 minutes a year. That would be pretty tough to live
> up to without redundancy built in to the circuits and access devices.
>
> -----Original Message-----
> From: jim.fitzpatrick@verizon.com [mailto:jim.fitzpatrick@verizon.com]
> Sent: Friday, December 19, 2003 1:07 PM
> To: ccielab@groupstudy.com
> Subject: OT: WAN LAN SLA Statistics
>
>
> Does anyone have any details regarding industry standard SLA percentages
> for WAN and LAN devices/circuits. It seems that everyone wants 99.999%
> uptime but that seems rare. I am trying to find some statistics and
> guidelines on more realistic figures. Any ideas on what should be an
> acceptable figure???
In the real world, it's silly to state some availability objective in
isolation. A rational design will consider the costs of downtime times the
probability of outage, and see if the resulting weighted risk in monetary terms
is worth further protection (i.e., more resources). There's often a point
where it's more rational to buy catastrophic incident insurance rather than try
to make the network more fault tolerant.
Pure availability often can be misleading, with a more important statistic than
mean time between failures being mean time to repair. If your data center is
in a remote location where it might take 3 days to get spare parts, it's
rational to go for more fault tolerance than something next to the Cisco factor.
There's a point in high availability where you reach diminishing returns trying
to make one installation, or network connection, more highly available.
Network fault tolerance methods aren't going to protect against catastrophes,
be they natural or man-made. Data centers in California are exposed to
earthquakes, data centers in Miami to hurricanes, etc. The military long ago
realized this and stopped building superhardened command posts like Cheyenne
Mountain, and began dispersing command posts -- even to the extent of having
airborne or other mobile facilities.
There can be unexpected hazards. While many New York financial institutions had
diesels and a week of fuel that they thought would protect them on 9/11, many
generators failed within 48 hours because the air filters clogged with the
unprecedented amount of dust.
This archive was generated by hypermail 2.1.4 : Sat Jan 03 2004 - 08:25:43 GMT-3