Share what you know with millions of people
Focus is the best place to turn what you know into remarkable content
0
If I need 5 9's uptime do I have to use a premise solution or can I go with hosted/managed solution?
If my requirements are to have 5 9's uptime then can I reach that level of support with an integrated premise based solution with updated hardware. Or is a hosted/managed solution the only provider that can guarantee the high availability?
Events
- Dos and Don'ts of Small Business Marketing May 29 @ 11 am PT
- Lead Nurturing 202: The Next Generation May 31 @ 11 am PT
- The Tricks to Paid Media June 6 @ 11 am PT
- Display Advertising for Brand Awareness June 20 @ 11 am PT
Related Activity
6 Answers
I guess everyone thinks they already know the answer to this, but I think everyone's answer will be different: what is 5 9's uptime? It sounds like the answer should be easy, and it's certainly easy to come up with an answer, but you'll find that vendors and customers differ over it.
So here's a definition: the service will be available 99.999% of the time. Meaning dial tone will be available, acceptable call quality will be available, and all call services will be available (for example mid-call actions like entering DTMF, transfering the call, etc). Maintenance periods don't count.
This definition seems reasonable, but actually it's extremely stretching. If a call server fails over whilst you're on a call you may not notice (assuming it's VoIP and the media doesn't go through it). But you may lose the ability to perform any further actions on that call, because some or all of your call context may have been lost. If a network re-routes, you may lose the call but be able to immediately re-dial. Or in the same situation you may suffer poor voice quality for a period of time, again re-dialling may fix it. What is acceptable service here differs between customers.
Solutions to provide all of the above 99.999% of the time are feasible but horrendously costly. I doubt many customers need it. If the service were to go down for several hours at night, most people wouldn't notice. So why pay for that level of availability?
To do this on-prem (presumably) takes the responsibility of the whole thing into your organisation. To do this via a hosted solution off-loads the problem. Whether hosted or on-prem is better depends on your view of life, but equally resilient solutions can be created with both. The maths that Phillip talks about doesn't care where kit or wires are located, only about the reliability.
Which raises the important topic of calculated reliability (based upon observed reliability of individual elements over a period of time, which is not always as useful as you would think) vs actual reliability (ie: what *you* actually observe in *your* environment). They're not the same thing, and it's the latter that really matters.
So here's the main point. Customers and vendors need to build solutions according to the actual needs rather than trying to meet mathematical targets. Don't just compare the number of 9's in the brochure.
www.mike-barnes.co.uk
You can certainly achieve 5 9's (Or Better) with a premise based solution if it is designed correctly. For years Avaya has prided itself on being able to provide 5 9's of reliability with it's S8700 (Now 8800) series of servers. The software is capable of mirroring processor functions, providing alternate gatekeepers, supporting multiple network interfaces and ensuring survivability at remote and central locations. Calls can automatically be re-routed to an alternate location via their remote survivable processor technology. I don't know of another PBX player today (not even Cisco) that can re-route calls at the blink of an eye without the caller knowing it.
With that said, there are some hosted solution providers that are able to achieve 5 9's by providing multiple data centers in order to do "disaster redirect". Couple that with having someone else managing your telecom 24/7 across multiple NOC's and hosted can certainly make sense while providing high reliability.
Sorry, Derek. You actually didn't answer the question . . . The answer to the first part is YES you can achieve 5-9's operational performance with a premise based solution. Oh, by the way, the 8800 is the first pass to integrate to the CS1000 platform and Avaya MMA is on its way out the door . . . (Thanks to the Telecom Dieties for the kindness and wisdom of this) . . . Sorry, I digress.
Actually, I find if you are using a hosted ONLY solution, you never reach 5-9's of performance. It's one of these where you have to remember basic math and statistical interpretation. Yes, the carrier side (network) is 5-9's and all the H/W is NEBS 3 compliant (5-9's). That only gives a 4-9's environment. Argue with the Mathematicians, not me. Operationally, the solution may achieve 5-9's performance, many networks do; that doesn't change the fact it IS a 4-9's environment, AND please, verify with IEEE, there is no such thing as a 5-9's Ethernet network . . .
Now, regarding what is the most efficient (Notice, I am not saying cost effective) solution is what I call a semi-hosted solution. This where the core of the solution is controlled and maintained off premise, just as Derek refers. In truth, that's basically how things have been done from a carrier perspective since the inception of telephony networks. The difference here being, the service provider provides an on-site appliance that acts as the interface from the customer site to the service provider's network . . . which gets to handle all the more magical and wonderful things our technology allows us to do.
Now, as to why a hosted service provider needs to have that on-site appliance. Heaven forbid, there is a network failure or an on-site emergency; that on-site appliance can should be a connect point for emergency services or as a inbound bypass for telephony services. In many states, they require ring back / lockup and locator services where the caller is identified within 200 feet and, at a minimum, floor . . . Hosted solutions have a difficult time with this and the database maintenance has never been the best.
I do have a sort of split point on advising the direction to go. If you are a large Enterprise, bluntly buy the system and invest in the people to run it. Remember this, make sure the people running the show have a Telephony background, not IT only. It's a white paper in itself to explain why. If you are in that small to medium size (25-400 stations), there are several good hosting companies. Just make sure there IS an on-site appliance as I recommend. For that SOHO (4-25), I'll be honest: Don't be afraid to look at an Astrick's solution, or even MagicJack and Vonage. They work. It all depends on how tech savvy you are and what telephony feature set you need . . . Good Luck
Christopher,
Regardless of the stated reliability of a provider or designed solution, you increase your 9's with the amount of the solution you can control. A premise based solution gives you more opportunity to to provide 9-building reliability with tools such as N+1 failover, copper line or other voice/data service redundancy, Emergency Trunk Overflow, UPSs on all critical [carrier and your own] equipment and backup equipment to minimize the inevitable downtime.
If you look at most hosted SLA's, if they do not hit the stated level of reliability their liability is limited to the cost of the phone service lost in an outage, which in most cases will not go anywhere near to covering the business loss caused by an outage. Controlling the system gives you the best way to control reliability.
Great question - and it brings up some larger ones.
Either a hosted or a premise OR a hybrid (premise+ hosted, like our Star2Star) can achieve 99.999% uptime. The important question is "at what resource cost?".
Single carrier connections are generally the weakest point in the chain between your telephony endpoint (handset, messaging system, call center) and your customer. There are many ways to mitigate the risk of single-link connectivity, from bonding multiple circuit together to backup routing to failover systems.
SynerTel designs and sell highly reliable premise (Toshiba, Allworx, Avaya, etc.) solutions as well as pure hosted and hybrid (where a small onsite appliance monitors hosted VOIP quality, packet loss, etc. and can adaptively alter the packet routing - or even trigger call re-routing to failover destinations such as cell phones, other branch offices, etc. in the event of a connectivity outage.
We find the key to creating a cost-effective yet sufficiently reliable solution is to start by understanding the profile of the business:
- where are its employees - centralized, distributed, mobile?
- is there a 'center of gravity'?
- what connectivity is currently in place? Are there term contracts restricting our flexibility to reshape/optimize it?
- what does the business continuity / disaster recovery plan look like? (Does one even exist? if not, we'll help craft one).
When this basic picture is sketched out, then its easier to 'run the numbers' for total cost of ownership and expected uptime for different types of solutions ... and the client firm can make a comfortably informed decision.
I agree with Mike. Building to a mathematical need as opposed to a business need is folly. There are a lot of variables that go into the equation. You can take the approach of controlling everything, as prem-based providers will say, but you can't control everything. It is not always feasible to build your own servers, data centers, carrier networks, PBX system, routers and all the other stuff. You are going to have to rely on them doing their jobs, working well together and not having bugs.
Find out from the customer what is important to them. Is inbound calling the most important, then focus on that. Is their call center most important, work on that.
The juice is not always worth the squeeze. Each business needs to know what is important to them, "EVERYTHING" or "I want 5 9s" are lazy answers. It is an easy check-off in a project list, but useless and expensive.
Build the requirements list, align them with scenarios, put plans in place and reaction plans for error that could happen. Then communicate, communicate, communicate.
Answer This Question