Home » Skype for Business » Skype for Business–Survivable Branch Appliance or Cloud PBX?

Skype for Business–Survivable Branch Appliance or Cloud PBX?

Skype for Business Server has been out in the general market for 14 months so far. However, one particular role seems to be dragging it’s heals towards release, and that is the Skype for Business Survivable Branch Appliance. If you are considering an SBA, then the likes of Sonus and Audiocodes have an SBA product. However, if you look under the bonnet, this will be a Windows Server 2012 (in the latest release or Windows Server 2008 R2 in older image versions) with Lync Server 2013 core components installed to act as the SBA.

There appears to be something in the pipeline to bring the SBA to full Skype for Business edition, but there seems to be no urgency to release this to the market. Why?..

Well Microsoft released Cloud PBX late last year, and have made great strides in rolling this out globally across all their Office 365 datacenters and media networks. It is clear, that this is the where their main energy, focus and direction are taking, which is a good thing! But it also begs the question over the viability of not only Skype for Business Server in the future, but the role of the SBA in todays’ ecosystem.

The purpose of the SBA was always to give some level of PSTN calling capability to a site when the connection breaks to the central site where the Skype for Business Front End Pool is located. These were ideal for long distance branch sites with poor inter-site connectivity that was prone to connection issues. Admins could guarantee a level of service to users at that site in order to provide the basic tools for “normal” business operation (Normal = basic toolset).

However, SBAs have the same problem regardless of the vendor you have chosen. If the connection to the front end server pool fails, users will lose the majority of their Skype for Business communication modalities, such as:

  • Presence
  • Instant Messaging
  • Peer-to-Peer Calling (outside of branch)
  • Federation
  • Agent functionality in response group
  • Delegate ringing (if delegate and delegator are in different sites)
  • Team calling (if team members are located in different sites)
  • Conferencing – except dial-in conferencing
  • Call Park
  • Call Forwarding updating
  • E-911 calling

As you can see from this list, the impacts to users when a WAN connection fails is significant, and therefore, ultimately has a major part in the decision making process around whether an SBA is worth the investment at a site or not.

Usually the decision is based on the predominant user role placed at the site for the business. For instance most businesses tier their staff into three main categories:

1. Front line

2. Operations

3. Back Office

Each category have their own critical dependencies in order to maintain business operation. For instance, front line staff are more than likely the revenue drivers within the business cog wheel, comprising of Sales staff, receptionists, and first line support staff. These users primary focus is the public face of the company, and the point of contact for the majority of transactions. Staff categorised as Operations could include, second-third line support teams, logistics, project office etc. Back Office teams are usually the people employed to oversee the internal operations of the business such as HR, Marketing, Estates, Admin etc.

Each of these tiers have different tolerances on system failure. In terms of Voice and UC, the company would be seriously handicapped if their sales team could not make phone calls to customers, and therefore, PSTN survivability is paramount. Whereas, the back office and to some extent the operations tier would be less impacted by a PSTN outage.

Note: The figures used in this post are subjective and not based upon any official research or survey. The figures used are to demonstrate the formula and considerations for weighing up the total cost of a Survivable Branch Appliance vs Cloud PBX in an ideal world.

As SBAs are quite expensive (around £6,000 to £20,000 depending on functionality & licencing) it is more than likely that they will be used for front line branch sites than anywhere else. For sites that depend on inbound routing (e.g. inbound sales, tech support etc.), they have a dependency of response groups being available in order to complete their role. In these sites, more than likely you will be deploying at least a Standard Edition Skype for Business Server along with a PSTN gateway.

In my experience, customers will only entertain SBAs as a viable solution if it is determined that the business would be penalised in a failure scenario. I have spoken with customers in the past who invested in multiple SBAs across sites where really they weren’t of any benefit other than nice to have. In my opinion it is a simple calculation, Cost to Business in Voice Outage / Cost of an SBA if the SBA comes out cheaper, deploy and SBA, if not, talk about other mitigations that are less expensive (deploy Skype for Business mobility for instance).

However, we now have Cloud PBX and all of a sudden we have some functionality we can utilise that could possibly negate the need for an SBA appliance all together. Perhaps (more than likely) this is the reason why Microsoft are dragging their heals with the SBA version..

Could we use Cloud PBX as an Alternative?

Cloud PBX has two main consumption tracks. Cloud PBX with On-Premises PSTN Calling (aka Hybrid Voice) and Cloud PBX with PSTN Calling (aka Cloud Voice). The main difference between each consumption track is the level of integration a business requires between cloud services and on-premises infrastructure, in particular the PSTN.

So which one is the solution to your SBA worries? The answer is both!

Now before I go any further here, there are three pre-requisites to all of this:

  • The branch site must have it’s own independent on site Internet connection
  • DNS must be pointing to the Public side of the central site Edge server pools. This protects the branch site from internal WAN outages, plus in distant remote sites (especially Eastern countries) where private WANs are hard to come by, or significantly slower than local internet, this could be an advantage!
  • The country in which the branch site is located MUST NOT restrict toll bypass

In probably 80% of cases, customers will be using a shared internet connection that peers out to the internet at a central datacenter. So this is an important topic to discuss right at the beginning when suggesting a Cloud PBX alternative. Customers may chose to invest in local Internet breakout specifically for Skype for Business Online as this will be invariably cheaper than WAN upgrades, or the cost of an SBA. Others may be more conservative because this does not fit their governance requirements (more than like .gov establishments). While this post is not going to enter into politics of such discussions, let’s assume the customer is open to the idea of deploying local internet facilities, or the site has one already!

So, what features do you get with Cloud PBX when the WAN link fails?

  • Instant Messaging & Presence
  • Team Call
  • Delegate Ring
  • Call Forwarding, Transfer, Simultaneous Ring
  • Peer to Peer Voice
  • Enterprise Voice
  • Conferencing (full suite)
  • Federation

As you can see the branch site can continue to operate at full modality support in the event of a corporate WAN link failure, compared with limited features if an SBA was deployed instead.

By leveraging hybrid between On-Premises Skype for Business server and Skype for Business Online with Cloud PBX, businesses can rapidly deploy site survivability with significantly less investment and complexity. Yes homing users in Skype for Business Online has some limitations, mainly around Response Groups at present. However, if your site was dependent on RGS, you would be deploying a On-Prem solution to that site in a traditional design anyway.

For a branch site with 50 people the cost of deploying an SBA would be around £6,000 for SIP only + annual software support @ 12% + Professional Services @ around £2,000 + 20 SIP channels to a ITSP @ £10 per channel per month + approx. £12 per user per month in call costs= £18,600 in year 1 and £10,800 per year thereafter. So over a 3 year TCO plan = £40,200 or £376 per user not including internal support services.

So what about Cloud Only Voice?

As you are probably aware, we have a choice of using on-prem PSTN gateways or Microsoft PSTN gateways to make / receive enterprise voice calls. Decisions need to be made about which model is right for the customer’s branch. We can start do this by answering the following question

Does Microsoft currently have PSTN calling available in the site’s country?

If the answer to this question is yes, then you have a valid reason to choose this deployment over the hybrid model. The next question on this decision is, does the customer need to port numbers to Microsoft, or use Microsoft’s own DDI pool? And then, the most important question, what is the cost of this model? This will be worked out as Qualifying Enterprise Licence + Cloud PBX + Calling Plan = £TOTAL per user

While cost will be a high priority discussion point, it should not be the sole decision maker in whether to use cloud voice. In order to achieve an accurate cost comparison here we need to consider a few scenarios.

Firstly, and perhaps most importantly, lets compare the survivability options for hybrid voice.

Hybrid voice allows customers to leverage their existing PSTN connections with Skype for Business Online users. This on the face value seems the most attractive solution to customers. They don’t need to port numbers and they don’t need to buy additional licences beyond Cloud PBX. All true of course, but we are still dependent on some on-premises hardware solution, what happens if the front end pool fails?

Sure, the customer may have configured a DR scenario and while there would be some level of downtime perhaps for voice calling to Skype for Business Online users, it probably wouldn’t be too much of an issue. But a more serious issue will arise if the PSTN gateway failed that prevented any calling for hours.

Forgetting failure for now and moving towards existing architecture, or sizing new architecture at the main datacenter. By employing the hybrid voice and moving branch users to Online, we instantly create more dependency and capacity requirements to the centralised architecture. All of a sudden, the SBC needs to support additional SIP sessions, the Skype for Business mediation server needs to handle more transcoding and encryption/decryption, and the Edge servers need to handle more media traffic (when WAN is out). Additional investments in hardware and licencing may be required here in order to beef up the centralised pool to handle this workload. So this needs to be taken into consideration when comparing costs.

Hybrid voice in this scenario would also increase the bandwidth dependency on the inter-site WAN link between central site and branch site. Depending on the route and limitations, several connections may need to be improved, even hardware replaced to support the media traffic between branch client and mediation server. Again, add this to your cost of hybrid voice.

Of course, you can prevent media over the WAN using internal firewalling and force the media over the branch internet and via the central site Edge. However, bear in mind that Edge servers are sized to handle only 30% external media calls and also by enforcing this, the best codec branch users will leverage will be RTAudio which is typically 50% less quality than G.711.

Now consider the administrative overhead to support and maintain hybrid voice. During a P1 trouble ticket, you are likely to engage several support technicians at some point from various skillsets, and no doubt they will involve the following teams:

  • Telephony
  • Server
  • Applications
  • Networks
  • Security

The incident may pass through sub teams within each team (level 2, 3 and maybe 4). All these support technicians have a cost to the business that can be divided up into billable hours and is relative to their experience. In addition, while troubleshooting the problem there are other costs associated that you may not have thought about, like:

  • Travel costs to site during troubleshooting perhaps
  • Expenses for field engineers if required
  • Cost of phone calls to suppliers, support desks, mobile etc. during troubleshooting
  • Administrative costs, ticket tracking, auditing etc.

Let’s take a look at a rough trouble ticket example:

Bruce experiences poor voice quality when making a call to a customer using Skype for Business Online Hybrid Voice. Symptom includes crackling, packet loss and sometimes complete failure to establish a dial tone.

So the first team Bruce is going to contact will be the IT Service Desk. Typically these people are not technical experts but can handle simple end user requests. The average salary of an IT Service Desk technician is approx. £17,000 per year or £8.72 per hour. Now the technician will be responsible for logging the call on the system, triaging with Bruce and if required to be escalated, be responsible for tracking of the ticket and chasing all parties for updates. On a support call like this, the technician will probably spend a total of 2 hours of their time from start to finish on this call. Cost of Technician £17.44

The technician can’t resolve the call, so they escalate to the Skype for Business Server (Applications) support team. Within this team there may be engineers who are varied in skill, so salaries will vary as well. Let’s assume the call is handed to the most skilled Skype for Business Administrator and their salary is £45,000 per year or £23.07 per hour. The Skype admin spends 4 hours troubleshooting with Bruce, collecting logs and analysing them, perhaps searching online for solutions, attempting fixes etc. But the problem is still present. The Skype admin has now billed £92.28 against the trouble ticket.

The Skype admin determines that the problem must be network related so passes the ticket to the networking team. Again, assuming the most skilled network admin is assigned, their salary is probably going to be somewhat similar to the Skype admin. The network admin spends 6 hours packet capturing at different network points between the branch site and the central site, checking configurations, QoS etc. and finds some issues which are fixed, but Bruce still suffers from the occasional failure to establish a dial tone. The network admin bills £138.42 against the trouble ticket and suspects there is a problem with a slave firewall between the sites, which may explain the intermittent call failure and engages the security team.

Typically, the security admin will command a salary of around £60,000 per year or £30.77 per hour. They spend 4 hours troublshooting the firewall and determined the firmware is out of date. After spending another 2 hours filling out change forms and applying the update, the problem still exists. The next step, the security admin calls the vendor for support and spends the next 3 hours (3p per minute = £5.40) on the phone troubleshooting with the vendor’s TAC support. TAC advise a component replacement (costing £800) at the branch site, so the security admin drives 2 hours to the branch site, replaces the component (1 hour) and drives 2 hours back (assuming in working hours time). On the way back they stop for some dinner and spend £20 on a meal. When calculating the mileage they travelled a total distance of 280 miles which is charged at 45p per mile = £126.00 in fuel and wear. Now totalling the cost of the security admins time and expenses the total billed to the trouble ticket is £582.18.

Bruce’s problem is now resolved and the company paid a total of £1,630.32! not including any financial costs suffered by Bruce not able to perform his job properly.

The majority of this cost was down to a problem with inter-site WAN connectivity, but in order to get there, the trouble ticket had to pass several hands and several points of diligent troubleshooting. If Bruce was enabled for Cloud Voice would the cost be any different? Let’s investigate…

So Bruce is going to contact will be the IT Service Desk. Service Desk triage the call and cannot determine root cause. They then raise a support case directly on the Office 365 Support portal. They also, assign the internal networks team to the ticket to rule out site connectivity.

The network admin spends 1 hour on the ticket and comes to the conclusion that there is no reason for the network to be root cause and informs the service desk.

The service desk spends time with Microsoft Support to troubleshoot the problem, spending around 6 hours of time collecting and passing information between Bruce and Microsoft. Microsoft discover an issue their end and resolve it. Service Desk now updates the ticket to resolved. The total time billed against the ticket from Service Desk is 8 hours @ £8.72 per hour = £69.76 plus the hour from networks @ £23.07 brings the total cost of the support ticket to £92.83

Ok, the problem with the firewall would still be present and a yet to be found issue, but it is less of a serious problem to business continuity because no service depends on the problem being fixed. Some people may argue that this is irrelevant, it’s a problem so it needs fixing, and I agree with you, but the reality is that if you go to your FD and say I need £1000 to replace this part, they are going to ask you what line of business feature is impacted? You will reply none, its a proactive fix. To which the FD will say “let me get back to you”, this means “lets leave it until it becomes a problem…” The point is it is not a critical spend, so from a business perspective it is an accepted risk.

The point I am trying to make here is that, by moving away from tradional on-prem solutions, in this particular case, an SBA deployment towards Cloud Voice solutions offer businesses substantial areas where money can be saved.

Going cloud voice is now becoming a more and more attractive solution… right, or is it?

Lets look at Cloud PBX with PSTN Calling from a price perspective. Let’s assume here that users within the branch are already consuming Enterprise E3 Office 365 licences. Here we have two options, purchase the Cloud PBX addon subscription and a calling plan, or upgrade E3 to E5 and purchase a calling plan. The base E3 licence is £14.70 per user per month, while E5 which includes Cloud PBX and PSTN Conferencing is £25.70 per user per month. In order to bolster E3 to include Cloud PBX and PSTN Conferencing £7.50 in addons is required (£5 for Cloud PBX, £2.50 for Conferencing) = £22.20 per user per month. Therefore, based on value E5 is the licence to purchase overall.

PSTN Calling plan in the UK is going to be around £13 per user per month for domestic calling and around £26 per user per month for domestic and international (guess).

Taking the licence cost difference between E3 and E5 + a domestic calling plan totals £24.30 per user per month or £43,740 TCO over three years in our 50 user branch site scenario. But wait… isn’t that more than the TCO of an SBA? Yes, by £3,540 in fact, but don’t forget the IT support service cost of each user per year. As a result of consuming a cloud only service, 50 users are going to place less of a burden on your IT support teams, the business is offloading the majority of support to Microsoft which is inclusive of the costs above which means the business can streamline and reduce costs in overall operations, and the need for ongoing investments in infrastructure in the same manner are no longer needed for these users. Based on around £900 per head per year IT infrastructure and services budget you can aim for a rebate of around 5% (£45) per user per year by leveraging Skype for Business Online instead of an SBA which is about £6,750 saving over three years. Which makes Cloud Voice in direct comparison a slightly cheaper option.

But then we don’t stop at just voice, or Skype for Business  either. By adopting Skype for Business Online as a complete UC solution and all the other benefits of Office 365 will see massive streamlining right across the business which improves collaboration, speeds up decision making, brings new capabilities to businesses that help them to expand more rapidly and explore new markets while reducing their capital overheads and providing a clear and achievable operation cost that can be scaled up or down depending on business demand and performance. Which all adds up to significant savings across the whole IT ecosystem.

The above scenario is of course a hypothetical best / worst case and by no means an example based on real world data. It is designed to make a point, that moving towards cloud solutions offers potential savings right across the IT landscape which should be considered when comparing costs with on premises. Let’s be honest, there are still concerns not to move to the cloud for some or all business IT elements and therefore, on-premises solutions may be the best for the customer. Ultimately, it is our responsibility to provide the solution that meets the customer’s requirements that is the SINGLE most important job of a consultant / architect, regardless of what technology leaders say or want you to do! Remember that!

I chose the more complicated argument to make this case, in comparing an SBA with Cloud PBX with PSTN Calling because it opens interesting avenues of discussion. However, for some customer’s, Cloud PBX with PSTN calling is not an option right now. Perhaps they are not ready, or perhaps they need time to gain confidence in the solution? So using Hybrid Voice gives these customers a taste of the possible, while still retaining a level of control they are comfortable with.

For this example, I am not going to go into great detail, but let’s assume a there is no need for a WAN investment. I need to buy 20 more SIP channels in my central deployment @ £10 per channel. I also need to buy 20 more SIP licences approx. £500 for the Session Border Controller in order to have enough capacity to take on the load of the branch site. Total investment over three years in this £2,460.

Now each user in the branch site will need the Cloud PBX addon subscription to their E3 plan. This is £5.00 per user per month + £2.50 for PSTN Conferencing, over three years for 50 users = £13,500

We will also need to budget £12 per month for PSTN calls for each user, so over three years this equates to £21,600

We are not going to be able to save much (if any) on IT Service support for these users, because I have the same networking if not more networking complexities to protect the business from, so I need the skills available, as Microsoft support involvement will be less than cloud voice. So let’s add the £45 per year per user support cost back on to these 50 users = £6,750

Total cost of ownership over three years for hybrid voice is approx. £44,310 (which is more expensive than cloud only voice).


From my own perspective, leveraging the use of the latest technology helps businesses embrace revolution. The days of the SBA are limited, as is on-prem Skype for Business server with SBC vendors now turning towards CCE appliances. For me Skype for Business Online with Cloud PBX is a viable alternative to an SBA and when calculated properly and customer requirements have been properly assessed can turn out to be the most cost effective solution overall.

It also provides an insight into Cloud possibilities allowing businesses to dip their toe in the water with Cloud PBX in real world scenarios to get real end user experience that will be critical to the decisions that lie ahead.

Clearly, the numbers game is close between the 3 possibilities and the real savings are going to be made in reducing IT Services and Infrastructure over time. So if you are looking for a quick win, instant return, this is not going to happen. Again, reiterating that the figures I have used have not been through any official ratification or analysis. These are just numbers used from different sources as well as some assumptions. Therefore, strongly recommend you conduct your own cost analysis into this before making a decision. Cloud solutions give businesses more scope to expand or shrink on demand in line with their revenue flow and also explore avenues of cost savings. Those savings, only you can estimate based on how you embrace the cloud..


  1. Hi Mark, thank you for this article.
    I’m sorry but not every Countries are like USA or UK. In Italy (where I live and work) SfB in Cloud could be a viable solution maybe in 2018, we’ve many remote site with local PSTN and a Sonus SBC 1000 with SBA is a good solution in that case, but the lack of SfB SBA is hard to explain to customers. I hope that MS take care of that Country and Customers around the world, not only who wants to move to Cloud.

    • Hi Luca

      Yes I agree, and this is why I said it is an option but not in all cases. Each case must be evaluated and the best solution for the customer recommended, that is the golden rule. For me it is an interesting argument and angle of a discussion point at least with the customer and community at large. Hopefully this article raises the question higher above the parapet. Perhaps the answer in some cases is Cloud PBX over SBA, in others SBA over Cloud PBX. Either way as long as the customers hard requirements are met and they are happy, then that is the main thing. That’s all that matters 🙂



  2. Thank you, Mark, for this great article. It’s helping me with some answers of my questions. I’m creating a concept for a solution without SBA/SBS and we have a Hybrid-Scenario here. The customer doesn’t trust the cloud only solution and in fact we don’t have the chance in the moment for this, because in Germany PSTN Calling is only in Preview. But we will deploy EV with the Hybrid without SBA/SBS first time and the other solution is coming soon, I’m sure;)
    Two questions: Can you please explain the reason of “DNS must be pointing to the Public side of the central site Edge server pools.”? And how can I do this for our centralized DNS without to affect the other brunch sites?

    • Hi Dennis. For this to work, the branch site must have it’s own local internet line, if it doesn’t then you’re not going to get survivability in this way. The reason you want to declare the public records is that you want to force the Skype client to connect to the SfB servers at the centralised Skype pool via the Internet and Edge servers, thus using the local internet breakout. To get the DNS pointing in the right way then you’d need to use an on-prem DNS Server in the local site to the internet public dns for the SIP domain. I am assuming that you’d have this at the site because if the WAN link went down, then you’d have no DNS services without one. Alternatively, if you don’t then you’d include public DNS servers in your DNS Server list, prioritising them below the the internal ones.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: