Do you know the difference between a VP3510 and a CH6010D? What about a UX4410 vs a GX6340? More importantly, do you understand Metaswitch’s hardware strategy and what it means for your existing and future network?
Before I start, I should warn you that this article could easily go off the rails. Here’s why.
- I started in Metaswitch support way-back in 2005. At that time the most widely deployed software level was V3.0.3 and Metaswitch didn’t yet have a voicemail platform.
- One of my early jobs was to sort through and catalog the closet where we kept all the US-based spare cards – there was some weird stuff in there.
- Years later, I was responsible for all Metaswitch hardware purchases and supplier contracts, and my team managed the warehouse full of stock hardware.
In other words, I could write a version of this article that documents in detail why you needed an AM1300 card to support GR-303 subscribers when the VP3510 chassis first came out, and goes into detail on the differences between a CB1000, CB3000, GX6320, RT6701, DM6710 and EM3110.
In fact, I may actually have written just such an article – which can be found somewhere in the Metaswitch archives in the basement under Cragmaw Castle.
However, since Ambien already exists, we’re going to take a different tack – and instead look at the big picture, and in particular try to understand what’s happening with the hardware platforms available today – and what’s likely to happen in the future.
So for your delight, I present: The Ultimate Metaswitch Hardware / Product Matrix: 2019 Edition. Or TUMHPM2019E for short.
But before we get into that, allow me a brief historical diversion.
Metaswitch hardware generations
- VP3500: Way back in the mists of time, there was the VP3500. It was used mostly for VoATM access lines and the last sales were in about 2004.
- VP3510: This softswitch used a compact PCI (cPCI) chassis, and ruled the Metaswitch world from 2004-2012. It was supported by a variety of rackable servers, which ran products like MetaView, the EAS voicemail system and the N-series application server.
- ATCA: Metaswitch introduced two new ATCA chassis in 2012 – a 2-slot version (6010) and a 14-slot version (6050). Now all the applications (from UMG to EAS to SAS) could run in the same chassis. And there was great rejoicing.
- COTS (Commodity Off-The-Shelf) servers: As Perimeta started to take off as an SBC it became apparent that the ATCA hardware platform wasn’t really necessary. It was an added cost and hassle, especially for some folks who wanted to buy only a Perimeta. So Metaswitch made Perimeta (and some supporting products) available on any random Intel server – e.g. Dell R600 series or even the cheaper R200 series. And if people wanted they could even bring their own hardware (BYOH) and just buy the software.
- Virtualization: Around the same time, Metaswitch was at the forefront of pushing for NFV – realizing that the telecoms network didn’t really need to be hardware-based at all, and that virtualization was the future. Therefore they released virtual versions of everything (except the UMG, which needs custom hardware for TDM connectivity).
- CX7000: In April 2019, Metaswitch announced a new hardware option, the CX7000 series. These provide a cheaper NEBS-compliant alternative to ATCA for everything except the UMG. And Perimeta (which is available on COTS Dell servers).
What does this mean for you?
In the sections below I’ll try to cover the most common scenarios that telcos may find themselves in, and hopefully offer some useful advice.
“I still have a VP3500”
No, you don’t. They’ve all been retired… I hope?
“I still have a 3510 or 2510 and a bunch of servers”
A lot of folks are in this situation. Although Metaswitch announced end-of-life on this hardware back in 2015 it’s expensive and time-consuming to upgrade, so I know a lot of folks haven’t started yet, or in some cases have no plans to upgrade their hardware. I mean, there are still DMS-100s out there and they were launched in 1979, so by that math maybe there will still be VP3510s in service in 2045?
If you really have no plans to upgrade then I should warn you that the last date of support on that hardware is November 30, 2023 – so you’ll need to have a plan to deal with any hardware failures after that date. And if you are hoarding spare cards please make sure you understand that there’s a big difference between a spare CPU card and a live CPU card…
If you’re considering an upgrade, then read this article about whether to choose ATCA or go virtual – although now that the CX7000 servers are available I would use those in place of ATCA whenever possible.
“I have some ATCA hardware – is that a problem?”
No. ATCA is still fully supported by Metaswitch, and will remain so for many years. No end-of-life announcement has been made and I don’t expect one for a few years at least. And even when it comes, there’ll be a very long lead time (it was 8 years from announcement until actual EOL for the 3510 platform).
On the other hand, one of the problems with all custom hardware platforms is that they will eventually end their lives, and it’s not always easy to get replacements, therefore requiring an upgrade of some kind.
By contrast, the key selling point of any commodity off-the-shelf hardware platform is that it’s very easy to acquire the hardware, and even if your specific server is EOL, you can easily replace it with the most recent generic equivalent – thereby upgrading to a much more powerful server at a bargain price.
So if you’re thinking very long term you should at least be asking yourself what your next hardware upgrade will be – and as part of that take into account that if you have TDM connectivity then the ATCA UMG is already 7 years old, and there have been no announcements about a replacement.
Will Metaswitch produce another UMG to support TDM in the future, or will everyone migrate to VoIP for all connections (as I wrote about here) before that becomes necessary? I don’t know the answer.
“I’m confused about the difference between CX7000 and VMware-on-Dell servers / vNow”
Yeah… if you read the fine-print, the CX7000 servers are actually running a minimalist version of VMware which then hosts a single Metaswitch application. So in a sense both CX7000 and vNow are both virtualized options.
However, in the CX7000 case you can basically ignore that detail. As a purchaser, you are buying a server from Metaswitch that fulfills a particular function (e.g. CFS) – one server with one product – and what’s going on inside that server, or even what hardware it’s running on isn’t really your problem. Metaswitch will support the product, and if there’s a hardware problem you’ll have a support contract with the manufacturer.
vNow is pretty similar – in that Metaswitch take ownership for setting up a little VMware deployment on some Dell servers that you buy from them, but in the vNow case you have multiple products (VMs) on each server – so three servers might make up a whole deployment – whereas with the CX7000 you’ll have one server per product, just like you have one GX blade per product in ATCA.
If you’re running VMware on servers you purchased yourself then all Metaswitch has provided you is the software, and support for that software – you are responsible for the virtual environment and all the hardware.
I know that’s confusing – fundamentally it comes down to how much responsibility you want to take:
- If you want the vendor to own their product, go with the CX7000 option.
- If you want to dip your toe into VMware but don’t really know how to do it yourself yet, go with vNow.
- If you feel comfortable with VMware and virtual applications then just buy the virtual products and set up and manage the environment yourself.
“I’m considering buying a Metaswitch and/or upgrading my system, what should I do?”
Like any consultant, my favorite answer is “it depends”.
But you’re in luck – because in this case I’ve produced an diagram showing what it depends on, and offering some simple questions to help you make a decision.
“What will Metaswitch do next?”
Frankly, I don’t have a clue, but if I was in charge of their platform strategy this is how I’d be thinking.
- Our core strength is in software, so the more we can focus on software instead of hardware the better.
- Therefore virtualization is great, and for those who aren’t comfortable running their own virtual environment, we should run our software on generic servers that are supported by someone else and easily replaceable. [This thinking seems to match their current strategy.]
- The one fly in the ointment is the UMG. As long as people need TDM connectivity they need a media gateway on dedicated custom hardware. For now the ATCA UMG is great, but eventually that hardware will reach the end of its life.
- The long-term options for UMG are (a) industry lobbying to move everyone to SIP and kill SS7, (b) find ways to sustain the life of the ATCA UMG hardware as long as possible, (c) develop a replacement UMG when the ATCA version starts getting old, or (d) partner with another vendor who has an H.248 media gateway that could be managed by a Metaswitch CFS.
As is clear from the wide array of options at the end here, I don’t know what’s going to happen – but the good news is that the ATCA UMG has many years of life left in it.
So if I were responsible for their UMG strategy, I’d be postponing this decision for a few years and hoping it became someone else’s problem!