LinkBack URL
About LinkBacksKeep a couple of multi-purpose spares and then replace when they fail.Originally Posted by localzuk
Keep a couple of multi-purpose spares and then replace when they fail.
A switch here never gets replaced until it dies.
I still have 2 spare switches, that get used to bring the network back to life, and as soon as the new switch comes its fitted.
In the land of milk and honey ... on mission critical services that had pots of money I could schedule replacements to fit my budget.
But in reality in schools this is not normally the case.
At the moment we've got a brand new network of Cisco switches, but I'd be planning to replace them in 5-7 years. Main need for replacement would be to increase bandwidth, or add additional functionality such as newer forms of PoE etc.
When kit goes in you plan for the life expectancy of it, including how long it is going to be out of warranty, whether you can purchase extended warranties (which you should be able to with that kit) and also if you have any idea of the lifecycle of the OS on it. You know how much it costs to buy the kit so you spread the money you squirrel away over the number of years you expect it to live.
In reality we know that folk don't carry this money forward but they can plan for when they have to find the funds and prioritise certain things. It does get a little more complex when you find that you need to replace the equipment due to increased requirements and so you have to look at how much of the investment has been used so far (e.g. you are 3 years into a 6 year life expectancy) and work out if you can make use of the kit elsewhere, etc ... the usual sort of moving stuff around really.
When kitting out a new ICT Suite and working out how much it all costs then the infrastructure costs should be worked in as well. A desktop PC might cost £500 to get from Supplier X but the software adds a cost, as does the network infrastructure ... you know that you will replace the desktop after x years ... and so factor that into your budget, but also do the same with the infrastructure.
basically, i have 2 things that make me lose sleep at night... number 1 is the failure of our primary DC, that would be a bad monday morning... and one of our 4, core switches dieing, that would be a 'why don't we just all go home because we are screwed till tomorrow' scenario as we don't have a backup, and we don't have a spare. it's a procurve 5406 with a crazy amount of routing in it. we could plug up a cascade of other stuff and the main site would sort of work, but our primary school would be completely without dns/dhcp/cmis/internet untill we got it back. Its under a rolling 12month warranty/replacement contract with the supplier, but the supplier doesn't keep the monsters in stock.
replacement is only likely when the whole shooting match starts getting unreliable to the point that it inconveniences those who hold the purse strings
I work to a 5-6yr rolling replacement. A switch shouldn't be out of warranty after 3 yrs. Sometimes we upgrade before end of life as it gives us a chance to move out out core or near edge switch further out into the network.
This follows my belief that the network is the core of our infrastructure. It is the single most important piece of IT equipment in the building, more important than the MIS, the email system, the broadband and the active directory. without a reliable network there is no MIS, AD, internet, anything
Our next purchase it to replace 4X 5500G's with 4x HP A5800's (with multiple 10Gigabit links).
This will allow us to have 40Gigabit connections to the blade centre (opposed to 6Gb/s shared between 12 servers). We'll then use the 5500's as near edge switches, complete with 10gigabit uplinks.
We use HP Procurve switches here too and have done for the last 5 or 6 years.
No need to have used the Lifetime warranty as yet and have spare switches that can be configured with the LAN configuration and dropped in where necessary.
I even have a spare Core Chassis that I picked up from ebay relatively cheap and if our existing one was to fail, I could easily swap the modules to get us up and running again in an hour or two.
Ours tend to be replaced as they fail and if and when they are needed to to support various things i.e. VoIP, Wireless.
I tried to a rolling programme but it just would take to long to do and funds were too tight, we then had an urgent need to get VLANs across the site PDQ for a campus wide BYOD program to go with our campus wide Ruckus, so I managed to secure funding to replace all the switches at once, great in some respects but not in others. I expect this lot to last a minimum of 5 years, its Juniper and all lifetime warranty stuff so should be pretty bulletproofI would love a rolling programme but it might take a long time to catch fully up / do them all when you have about 30 switches and when buying big name L3 switches you need a few £K per switch and have a very small budget.
the 3750E/G is a very useful switch to use as a core. Earlier 3750G's and later 3750E's can be intermixed in a redundant stack configuration although negotiating down to the slowest stack technology i think. Configuration is effectively shared among members as a new stack master would be promoted should there be that kind of failure....Would keep one or two spares of those for sure.
Would also keep spares for the edge aswell, have used all of the models of switch you've mentioned and never had any reliability issues. The only think have had fail in seven years is a shortwave sfp gbic which died.
It may be worth speaking to a cisco partner regarding what kind of hardware coverage they can provide for your inventory of cisco gear.
The answer is when it's broken or when there is an excellent reason to do so.
This year will see me complete the program of replacement that I started 5 years ago.
The replacement program was to replace all of the unmanaged switches that the LA specified when they managed the school.
Keeping spares of £3k - £5k switches wouldn't go down well in many schools... The edge switches are a mix of PoE, 48 port, 24 port etc... So, we'd need to get spares of the best model we use to ensure we have coverage there, and again, they're £2k+
Speaking to a cisco partner is a good idea though, will do that Monday.
When they die - although I did sneak in a new gig switch recently as we had run out of capacity
I'd be moving them out to the edge. If you really need to replace a core switch then you have a spare that can handle it.
I am quite perplexed at the number of posters who replace 'when they break'.
I'm honestly wanting to know if the same people use this method with their SIMS, AD and other critical services? Is this strategy particular to network equipment or does everything gets replaced when it breaks? If the former, then why regard network infrastructure as second rate to server infrastructure - when I regard the two with equal (or even more!) importance? just a sub question and sorry to hijack
yes, i too am quite surprised by that response. i can only guess that because further out to the edge you go the more your likely to find cheap and cheerful, cheaper to replace switches. But as you say, it still means serious disruption to those on that edge.
the previous point about cost of keeping cisco spares, it's a fair point. They aren't cheap devices to keep in the stock cupboard, but then you pays your money (or in this case the msp did). That's what you get when you go cisco, not to mention some killer education discounts, or atleast used to be the case. The cost of maintaining spares and/or smartnet agreements is part and parcel. it's an extremely favourable position to be in to have that level of kit from core to edge.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
