Smartnet - should we purchase?
Chris
Cisco switch. At our main site we have 3 routers and 2 switches for a
total of 13 devices. Our vendor is looking for 5k to up the Smartnet
for another year on all devices. Is this worth it? Would I be better
to save the money each year, and just replace a device in the event it
fails? I was thinking of just renewing on the 3 routers in our main
location, as one of them failing could take down the entire network.
Any thoughts or suggestions? Thanks.
Chris.
Walter Roberson
Chris <chris_c...@forkliftsinc.com> wrote:
:We have 5 remote sites, each with a 2621XM router and a 24 or 48 port
:Cisco switch. At our main site we have 3 routers and 2 switches for a
:total of 13 devices. Our vendor is looking for 5k to up the Smartnet
:for another year on all devices.
Let's see... a 2621XM is SmartNet Category 5.
You don't indicate which models of switches you are using, so I'll
have to conjecture. As a 2621XM is not exactly the fastest router on
the block, my suspicion would be that those switches are probably
somewhere around the 2900XL series. But the 2900XL appears to be EOL
and you do not appear to be able to get SmartNet at all for those,
with the sole exceptions of the 2948G-L3 and 2980G, both of which
are much faster devices than the 2621XM, and the 2980G is gigabit
fibre oriented. There was no 2924G-L3, and I suspect you wouldn't
deploy radically different switches depending on the port count.
If you are in the new 2950 series, then as you likely wouldn't need
advanced features, then it appears you'd be Category 1 for 24 ports
and Category 3 for 48 ports.
If you are using a modular 4000/4500 series switch just for a 24 or 48 port
configuration with a router as slow as the 2621XM, then you probably
should have gone for a Nortel device instead: Nortel has some solid
switches suitable for smaller wiring closets.
I'll estimate, then, that for your remote sites you would be in for
5 x PKG5 (router) + 3 x PKG1 (24 port switches) + 2 x PKG3 (48 port)
For your main site, you haven't given us a clue as to what routers
you have, but if you went 2621XM for the remote sites and you have
3 routers, I'll hazard a guess that you have something like 2 x 2621XM,
1 x 2651XM (category 5 too), and, hmmm, say 1 x 2950 24 port +
1 x 2950 48 port.
The total under these shaky conjectures would be
8 x PKG5 (router) + 4 x PKG1 (24 port) + 3 x PKG3 (48 port).
If we hypothesize a service level of CON-SNT (8x5xNBD) for each of these,
then we can put together some pricing. I've used shopper.com for price
estimates; costcentral.com has slightly lower prices than what I've used.
CON-SNT-PKG1 : 4 x $US 55 = $US 220
CON-SNT-PKG3 : 3 x $US 175 = $US 525
CON-SNT-PKG5 : 8 x $US 340 = $US 2720
--------
total $US 3465
Beyond that, either your vendor is not giving you the discounts you
could get, or you have a higher level of service, or you have different
equipment than I've hypothesized.
: Is this worth it? Would I be better
:to save the money each year, and just replace a device in the event it
:fails? I was thinking of just renewing on the 3 routers in our main
:location, as one of them failing could take down the entire network.
Looking at these figures, the prices would come out at about $US 400
or $US 520 per remote site, and for that price you get hardware and
software support. If they are old devices than the software support is
not going to be very useful. Also, for -some- models of the LAN switches,
the software updates are free.
If you were to buy a spare 2651XM (~$US 2500) and a spare
48 port 2950-48T-SI (~$US 1600), your one-time cost would be
about $US 4100, compared to recurring costs of about $US 3500 per year.
If, then, you are sure you will not need to do software updates on the
devices (a risky assumption), then going the spare-stock route would save
you money in about a year.
However, you must ask yourself about "opportunity costs".
Suppose one of the remote sites goes down one afternoon with a hardware
problem. Suppose further that it happens in time for you to reach the
courier and get your spare sent out overnight, in time to arrive around
noon at the remote site, and suppose again that you have someone at the
remote site that can be trusted to be able to pull the old device out
of line and put the new one in place. That'd be about $US 75 in
shipping, and about (say) 6 hours of daytime network downtime
(plus overnight downtime) at the remote site. How much business would
you not be able to carry out if that were to happen? If as little
as 10% of your remote workforce relies on that network connection
and essentially can't continue without it, then that's 3 or 5 (so call
it 4 on average) people lost for 6 hours. 4 people x 6 hours x $40/hour
of salary + benefits, plus the shipping, is more or less $US 1000
for that single incident, even assuming only 10% work stoppage -- multiply
accordingly for higher number of people unable to work, and add lots for
lost orders, inability to do remote backups or tally receipts,
or whatever.
Now consider the alternative: CON-SNTP service level gives 24x7x4
service. If your remote sites are within range of a Cisco depot, then
for a standard device, they are supposed to be able to provide a
replacement in about 4 hours, any time of the day. (I've never had reason
to test that in the middle of the night!). If we're still going with
the afternoon breakdown scenario, that'd be about 3 hours of worktime
lost, and no shipping charges, roughly halving the day-worker stoppage
cost, and mostly eliminating any after- 5pm network connection
opportunity costs. That takes the opportunity costs down to about
$US 500 {under the same assumptions as before}.
The cost of going for CON-SNTP is roughly 50% more than the cost of
going for CON-SNT. Interestingly, if one takes the cost estimates
I prepared above, that takes you to roughly $US 5000 per year,
which was the price you mentioned.
So the figures are close. A one-time charge for "hot-spares" is
about the same as the yearly maint cost -- but unless your
remote sites are pretty much in easy driving distance of each other,
your opportunity costs dominate the equations nearly completely.
At our main site, we have prepared estimates of the opportunity costs,
by examining salaries + benefits, dividing by working days per year,
estimating downtimes, estimating percentage of people who would be
unable to find something else to do for a few hours, and taking weighted
averages of the portion of our people who work particular hours
[our working day is flexible, and lasts more than 8 hours from the
time the first person arrives to the time most people have left.]
What we determined was that for us, a single incident would cost us
about $C 10000 (about $US 8500) even if repaired with all due dispatch.
(We aren't a business, so I didn't factor in any cost of lost sales.)
Our equipment is also a little higher end than what I conjectured you
have, so the price for us to run hot-spares just for our core equipment
would be about $US 25000. At roughly $US 5000 a year in SNTP SmartNet
contracts, it would take us *years* to catch up to the one-time
hot-spare costs.
On the other hand, for our endpoint wiring closets, we took the
opposite tactic: with ~20 identical switches in the same building
and software updates not an issue [no bells or whistles on the switches],
it was much cheaper for us to buy a couple of hot-spares. We've used
for exactly that purpose from time to time; the rest of the time,
they have helped out by providing test-benches or capacity for temporary
projects. We have saved quite a bit by using "self-maintenance" on
these devices, in circumstances where we are -right there-, only
a fraction of our people are affected by any one potential failure,
and we can have a replacement switch installed in 20 minutes.
--
"Infinity is like a stuffed walrus I can hold in the palm of my hand.
Don't do anything with infinity you wouldn't do with a stuffed walrus."
-- Dr. Fletcher, Va. Polytechnic Inst. and St. Univ.
Chris
So you know, we have 3550 series switches at all sites, (2) 48 ports at
the main site, (1) 48 at a remote, and (4) 24's at the other remotes.
All site have a 2621 XM (6), and the main site also has (1) PIX-515E
and (1) 2651 XM. All sites are within 2 hours of the main site, and
minimized downtime is critical as we are a business and loss of data
communication can result in loss of revenue. Does any of this change
you take on our situation?
Thanks for the independent viewpoint,
Chris.
Scooby
news:1102688241.4...@c13g2000cwb.googlegroups.com...
Ditto on that, Walter - incredible amount of information.
If minimized downtime is critical, I would say to buy some spare equipment.
You will get the equipment to the sites quicker than you can have it
delivered by Cisco. You can buy minimal service contracts for the software
updates and hardware replacement. In the long run, you'll save time and
money.
Jim
Walter Roberson
Chris <chris_c...@forkliftsinc.com> wrote:
:So you know, we have 3550 series switches at all sites, (2) 48 ports at
:the main site, (1) 48 at a remote, and (4) 24's at the other remotes.
:All site have a 2621 XM (6), and the main site also has (1) PIX-515E
:and (1) 2651 XM.
Ah, I was close. I see I got the 2621XM and 2651XM at the HQ right.
I had conjectured a second 2621XM there because you had indicated
that you had three Cisco routers: there aren't many people who
would call a PIX a router [though I do regularily argue the point
here that it isn't -not- a router either.]
Would the 3550's be WS-C3550-24-SMI and WS-C3550-48-SMI ?
Those are Category 3 and Category 4 respectively. 4 and 6 if they are
EMI. CON-SNT-PKG4 is about $US 250, or about $400 for the SNTP version.
The SmartNet part for the 515E is CON-*-PIX515ER for the restricted
bundle, CON-*-PIX515EUR for unrestricted, but that is the new-style
part number and you will have trouble finding a price for it outside
of New Zealand [for some reason, New Zealand vendors are quicker
to put in the new SmartNet part numbers; a couple of places in the UK
are next, and everywhere else lags way behind.] The old style
has the Restricted as Category 6, and Unrestricted at Category 9;
CON-SNT-PKG6 is about $US 450, and CON-SNT-PKG9 is about $US 820.
:All sites are within 2 hours of the main site, and
:minimized downtime is critical as we are a business and loss of data
:communication can result in loss of revenue. Does any of this change
:you take on our situation?
Yes, being ~2 hours away makes a noticable difference in
the strategies. Cisco does offer a 24x7x2 SmartNet, but that would
only get you about the same delivery time as you could do yourself,
and you have to be within 25 miles of the depot to be eligible.
A spare WS-C3550-48-SMI starts from about $US 2750 (lacc.com),
about $US 500 more from better-rated stores.
A spare PIX515E-R-BUN starts from $2225 (lacc.com again). If,
though, you are going to hot-spare that [which you only have one of],
then I would suggest that you might be in a situation where it would
make business sense to use a failover. The main 515E would have
to be Unrestricted for that (makes a big difference in the price),
but the failover unit, PIX-515E-FO-BUN, starts at about $US1875
(lacc.com once more.)
[I'm pulling these prices from shopping.com.]
I gave the price of a spare 2651XM in the previous posting.
Now, factors at work: the 3550 series is still very young, and there
are big software improvements and fixes still coming out for it. You
would want to be able to update your 3550 software. The PIX 515E is
not so young, and the 26x1XM are even older, but both still have very
active software development; on the PIX side, PIX 7.0 should be out any
day now (it's been in beta since early in the year.)
For the switches, you might consider software support only; that would
be CON-SAS-* for TAC access and minor release update eligibility, and
CON-SAU-* for SAS plus major release eligibility.
CON-SAS-PKG3 ($US 158) and CON-SAS-PKG4 ($US 233)
CON-SAU-PKG3 ($US 165) and CON-SAU-PKG4 ($US 250)
Looks like it's pretty much the same price as CON-SNT-*...
So... I'd suggest one hot spare 48 port switch (which
could be dropped in place of a 24 port switch as well),
one hot spare 2651XM (which can be dropped in in place of a 2621XM
as well), and then getting CON-SNT support on the switches
and routers. When you run into a problem, you drive the hot spare
out to the remote site [presuming you have someone who can
be delegated to do that on short notice], swap, and bring the
failing device back to HQ for testing. If it needs to be
replaced, you get it done under the NBD replacemnt terms of
the CON-SNT package. Then at the next opportunity, you
install the repaired or exchanged device at the remote site
and take back your hot-spare.
Question: as you have routers and fairly small port counts
at each of the remote sites, is there a factor you haven't
mentioned that led you to the 3550 multilayer series instead
of the 2950 layer 2 series? And if there was, is the difference
something you could do without for a few days? If a 2950 would
be good enough for a few days, you could reduce the hotspare
costs by having your hot spare be a 2950-48 SMI.
--
vi -- think of it as practice for the ROGUE Olympics!
BradReeseCom
You may want to explore "Cisco Board Level Repair" options in
combination with Hot Swaps and SMARTnet:
http://www.bradreese.com/cisco-big-iron-repair.htm
Additionally, the Cisco SMARTnet Pricing Calculator may be helpful:
http://www.bradreese.com/cisco-smartnet.htm
Sincerely,
Brad Reese
BradReese.Com Cisco Resource Center
United Kingdom: 44-20-70784294
U.S. Toll Free: 866-864-0506
International: 717-489-1521
Fax: 775-254-3558
AIM: BradReeseCom
Website: http://www.bradreese.com/contact-brad-reese.htm
Ted Mittelstaedt
"Chris" <chris_c...@forkliftsinc.com> wrote in message
news:1102688241.4...@c13g2000cwb.googlegroups.com...
Hi Chris,
Just another view of this issue for you.
For starters, if you review all hard downtime you have had over the past
couple
years I would bet that hardware failure of wan and network switches is
probably
at the bottom of the list. I would bet that the circuits that tie all this
together
are probably the first to go. If you prioritize the downtime causers in
your network
I will bet that there are other things that you could spend the service
money on
that would reduce downtime more.
Secondly, every business claims that a network outage is going to cost
them
revenue. But in reality this is seldom true. For example have you ever
called into
your doctors office to schedule an appointment and been told that they
couldn't
schedule it right now because the computers were down? Did this really
cause
you to immediately switch doctors? Of course not. So sure, your doctors
office
scheduling girls are going to dislike irate phone calls and such, but they
really
are not running a network where a day's outage is going to cost the doctor's
office
anything, now are they?
And consider also whether your business is 24x7 or not. If it is 9-5,
then you
have TWO THIRDS of each day that the network is not needed for customers.
If you have cold spares in hand, that is a whole 16 hours each day that you
can
go work on replacing a failed network device.
If I were in your shoes and I didn't have a 24x7 network I would drop
service on
everything except for the PIX. After all keep in mind the PIX was never
Cisco's
product to begin with, Cisco acquired it, it doesen't run IOS, and Cisco has
moved
their really cool stuff - the intellignet network stuff - into their
routers. As a firewall
the PIX really isn't that exciting anyway, it doesen't virus scan, you can't
run a
porno sniffer on it to NetNanny your network, etc. etc. Most of the modern
viruses and cracks go right past it. Service on it makes a lot of sense. I
would
also buy a 48 port switch and a 2651XM
and keep them boxed and use their boxes for a footrest under my desk.
One other thing about service. Cisco has gone to a policy where if you
drop
service on a device, then a year or so later pick it up again, you cannot
get hardware
coverage on the device, only software coverage. In other words they require
continuous coverage to maintain hardware replacement coverage. This was to
stop the jackasses who bought Ebay routers then put them under service then
returned them to Cisco 6 months later when they broke and it's a good policy
IMHO..
Ted
Walter Roberson
Ted Mittelstaedt <te...@toybox.placo.com> wrote:
: Just another view of this issue for you.
: For starters, if you review all hard downtime you have had over the past
:couple
:years I would bet that hardware failure of wan and network switches is
:probably
:at the bottom of the list. I would bet that the circuits that tie all this
:together
:are probably the first to go. If you prioritize the downtime causers in
:your network
:I will bet that there are other things that you could spend the service
:money on
:that would reduce downtime more.
Prioritizing the downtime causes is a good strategy.
As a datapoint, around here the #1 downtime cause would possibly
be bonehead mistakes by overworked system/network admins [i.e., me.]
A close #2 would be hardware failures of our (non-Cisco) core switch.
Fortunately my 3750G arrived yesterday so that old device will be
replaced ASAP.
#3 would be software issues with our distributed filesystem:
although in theory it can lose any node without difficulty, in practice
if the node is the current master node, there's about a 1 in 3 chance
that other nodes will be affected. [The master node itself is not
fragile, but it does also happen to be our backup server and we've
been having problems with our tape drives that have required
reconfigs and reboots.]
#4 would be hardware failure in the stacking module of one of our
(non-Cisco) switch stacks.
Hardware failures in our Cisco WAN routers would be way down the list.
I think we did have a daughterboard on an Cat5000 RSM go once.
Unscheduled downtime by our ISP has been relatively rare. They did make
a typo in the Cisco Blaster Mitigation procedure and that caused us a
couple of days of mysterious grief (they were blocking -every- 92 byte
packet instead of just 92 byte icmp echo packets.) And their Cisco 6500
series routers have lost the route to us a couple of times (though
certainly not recently.)
Urk, I nearly forgot one of the biggies: that #@#$ non-Cisco core
switch has $#@$ software, and I spent more than a few long nights
trying to make it work. That was before I realized that the best thing
to do was to set it up once as simply as we could get away with,
and then leave it alone. I can't say that that affected daytime
operations much, though: I worked on it after regular hours.
The other network admin and I have been planning how we are going
to take our revenge on that &!@# box.
: PIX
As a datapoint, our PIX 525 has never had a hardware problem, but I
have made good use of the software support, both upgrades and
reporting bugs.
Our 506E arrived with the power connector block pushed so far in
that we couldn't keep a power cord on it for very long.
The power connectors on the 501's are not very good; we had some
issues with vibrations eventually working them loose, until we
strapped them on.
--
Oh, to be a Blobel!