If youre trading out of oracle for sansay or sonus i dont suspect you're going to find any revelations. Its mostly going to be a lateral move for slightly better support and potential feature loss. The Acme is, despite oracles best efforts, still the best commercial SBC out there in this particular space. The ability to extend with HMR and Lua scripting makes it super powerful.

Some considerations after having run a 100k+ user geographically distributed network:

1. Whatever you buy, go virtual. This breaks the hardware lock-in cycle
   and removes the typical CPU binding of SBC's. Make sure it runs on
   COTS (and preferably plays nice in public cloud).
2. Deploy more, small instances (5-15k reg). Use deep SRV records (4+) .
3. Strongly consider a Opensips/Kamailio approach. Your up-front
   investment at your scale will be approx the same, spent in
   intelligent manpower over the hardware solution but your recurring
   will likely fall to a fraction.
    1. plus of this is if youre primarily UDP on the access side
       reasonable HA can be delivered.
    2. if you are using TCP signaling, TCP failover is not as good as a
       Acme or Sonus (no state sync).
    3. If you are using TLS the ability to automate certs with
       LetsEncrypt, vs not doing TCP state sync could be a wash.
    4. Ive been having lots of discussions with the devs on the
       opensips project about solving many of these issues in large
       scale carrier networks, and its certainly do-able.

I would suggest having a look at the Metaswitch Perimeta as well. It does the SBC thing as well as any and can run virtual.

I would also suggest a demo of the FraFOS ABC SBC (an absolutely atrocious name for a very solid piece of software).  I have personally run this one and was very happy with it. It also runs cloud native nicely.

The only rub here is if you need large scale transcode. FOSS isnt taken off the table but complexity goes up a little bit. The Frafos could be a solution here.

FWIW if/when you do this, I know a used Acme broker that might be interested in your cast-off gear as well.

Best of luck!

-Ryan Delgrosso

On 2/23/2018 12:49 AM, Jeff Anderson wrote:
Currently registered endpoints in the tens of thousands. It would take us a long time to grow to 100k but it’s possible as we do more and more legacy switch migrations.

I am not aware of anything too exotic but also not sure what those exotic feature might be. We are an ILEC in certain markets if that tells you anything. We are spread across several states and have class 5 switches in each market.

We plan to integrate with a routing engine as part of this initiative.

The incumbent is Oracle one in each market.

We like the idea of centralizing the signaling to certain markets and distributing the media which currently only is available with sansay.


On Thursday, February 22, 2018, Ryan Delgrosso <ryandelgro...@gmail.com <mailto:ryandelgro...@gmail.com>> wrote:

    Hey Jeff,

    Can you qualify your requirement more? Ive got personal experience
    with tons of solutions of various sizes, but some are a poor fit
    for certain shapes of organizations.

    Endpoints numbering in the thousands? tens of thousands? bigger?

    All registered access traffic or anything more exotic?

    Also who is the incumbent SBC?

    On 2/22/2018 10:32 AM, Jeff Anderson wrote:
    We are in the market for a new SBC platform and have narrowed our
    choice down to two different products. The Sansay VSXi or the
    Sonus SWe. We will be running on a vmware hypervisor in several
    different markets. The SBC would be for SIP access with
    Broadworks and SIP peering with other carriers.

    I was wondering if anyone can share with me any of their
    experiences good or bad with either company on or off the list.



    VoiceOps mailing list
    VoiceOps@voiceops.org <mailto:VoiceOps@voiceops.org>

VoiceOps mailing list

Reply via email to