Service assurance with the Cloud
Date: Tue, 02/14/2012 - 19:13
Third time unlucky? It is said that two thirds of mobile users will give up any new service after only two failed attempts to use it. In a culture of instant gratification, churn is the service providers’ nightmare - hence the need for service assurance (SA) for an optimal subscriber experience
Recognised standards for SA would do much to attract new users to the cloud. Rob believes further that service assurance will become even more important as individuals and companies come to rely on the cloud. Keeping everything going, safely and with suitable mitigation for inevitable failures will become a major concern.
What are the key assurance criteria? How can they be quantified? And by whom? Rob asks our panel at NetEvents EMEA Press Summit, Rome, for answers on which to build the future of business in the cloud.
Panellists: Marco Wanders, EVP & Head of Products Division, Acision; Kamal Okba, CEO, Mauritel; David Howorth, Regional Vice President of Cloud and IT services, Verizon Business; Tim Gigg, Head of Interna- tional Transport Programmes, Vodafone Group
Introduced and chaired by: Rob Bamforth, Principal Analyst, Communication, Collaboration and Convergence, Quocirca
Excellent, thank you. Good afternoon everybody. What a great slot, hot day, near Rome, wine country, and we've got an exciting subject, service assurance. We'll try and make it interesting and we shall try and cover lots of different aspects of service assurance in the cloud. So hopefully despite you having rich Italian food and hopefully a nice lunch we'd like to have a bit of a debate. We'd love to get questions from you all. And let's take it from there.
So my area of coverage is very broad, communications, collaboration, convergence with analyst house Quocirca. What I really wanted to do with starting off this session is to just let's think about what we're talking about here with cloud-based services and cloud-based infrastructures. An incredibly complex and mixed environment and what could possibly go wrong.
Well very largely we are in a hyperconnected networked -- heavily networked world in all aspects, not just the IT infrastructure but all of those other aspects how we communicate and link together. And there's the potential for massive unpredictability and uncertainty.
And when we talk about moving IT infrastructure, software and platforms into other places and delivering them as cloud-based services we're only increasing potentially some of this unpredictability and uncertainty. And yet, from a meeting I had last week with 20 or so CIOs at a dinner event in London, and because there was wine and so on they were fairly frank, open and honest with what they were saying, they said this for them is a risk. They understand the benefits, financial, business agility, flexibility, no problem. But personally that's a risk because they have to deliver some element of certainty in this uncertain and unpredictable world.
And it's so unpredictable that we're seeing I think a level of difficulty in order to understand exactly what might happen with services. So for example here I'm sure this hotel is perfectly happy with their WiFi service for most guests and most functions where they have the odd businessperson and lots of tourists. It's probably perfectly adequate. But here we have a big wave crashing of all you folk and that was something that perhaps they didn't expect, although they did have a booking so maybe they could predict a few of these things, but there's a level there of uncertainty and change which is very difficult to cope with. And yet services still have to be delivered.
Services have to be delivered because of something we've been saying for a very long time in this industry. I pinched my graphic there, anyone, anywhere, any place, any time or any time in anything, from a slide that I delivered in 1995 when we were talking about the Internet and Java and all sorts of smart services flooding onto networks on lots of smart devices. Guess what? We are pretty much there. We can deliver pretty much anything, anytime, anywhere to anyone on any device.
What we can't do that well is assure it to the specific service, the specific individual with specific levels of access. That's what service assurance is about, predictability, reliability, security, being able to do all of this safely and to be able to deliver all of those services. And why is it that we can't quite do that yet? Well maybe it's something to do with how we view values and especially the value of infrastructure. And there were a lot of conversations earlier on today and yesterday about how it might be possible to deliver fantastic bandwidth and connectivity virtually anywhere but who's willing to pay the price.
And when you look at what things cost and what people are willing to pay, here's a little sliding scale of what we all as consumers are essentially willing and able to pay.
You can see that sometimes technology doesn't feature quite so highly in our daily spending. And perhaps sometimes we just take it for granted.
Now if I'm a CEO or a CFO of a large organisation maybe I take it for granted.
Actually I've also heard from some of you that CEOs are pandered to and given extra levels of service because that's who they are and everybody else in the organisation gets something less. So they don't necessarily understand the difference.
So my question is what price is everybody willing to pay to get the level of service assurance that they need. So I've got with me today a panel, a distinguished panel of various different organisations and there is some consistency between them in terms of the conversation that we've all had earlier about how this issue needs to be addressed. And so I'd like each of them in turn to open up an aspect of this debate about service assurance and to just say from their perspective where they're coming from. So if we take in turn perhaps, Marco first from Acision. Marco?
Thank you Rob. So if you talk about service assurance in the cloud there is notable aspects and at a high level there's four, right. There's application performance which comes with (inaudible), availability, there is security of course and there is – and that's what I want to focus a bit on is what I call when or what (inaudible) network optimisation. Because in the end what you see is that security very important but there is many ways to solve that, guarantee that. Availability within the hosted or the cloud environment, obviously redundancy and this morning obviously the discussion about the role of Ethernet, speed etc. A lot of things that are quite complex but can be
managed in a more local environment.
If you talk about outside of it then it becomes unpredictable and uncontrollable.
Because think about everyone that wants to access some sort of services over the same access infrastructure. And it doesn't have to be a cloud service but if you have a mission critical enterprise cloud services and someone else tries -- and you try to access that over a mobile network and someone else tries to access a Youtube video and it's on the same cell and if there is hundreds of people on that same cell then all of a sudden your mission critical application might be hard to reach.
So the key question here is how can you guarantee that those mission critical applications can be reached at any time like Rob was saying earlier from any device and any location. And the busier it gets, and definitely in mobile environments, the harder it gets to access any type of data so let alone a mission critical application. So that means that in the access you need to start thinking about okay, what is the quality of experience that people need to sign up for. Because in the end, no matter how good the cloud service is, no matter how available the applications are, no matter how secure the applications are, if you cannot access it it's absolutely worthless.
So that is one of the topics I would like to spend a bit more time on.
I think from my side -- I work for Vodafone -- we're probably fairly embryonic in the cloud. We have a number of products which we're now taking to market. We've got one product which is One Net which is a SOHO/SME converged service. It's really PPX in the cloud or cloud telephony. We've got about 1.5m users on that. We're now looking at taking that out to the multinationals and one of the things we're pretty quickly realising is we need some form of parameters of how we deliver that service.
And simple things within Europe like latency where we have a data centre in Ratingen and another one in Dublin. When you're looking at Asia Pac, when you're looking at Africa make the problem a whole different solution and requirement that you have there.
So those are the things we're looking at. So from my side I'm looking at latency. We mentioned availability. I think we're also looking at a dynamic network. We want a customer experience. How do we do that with the cloud? It's fairly simple with an IP/MPLS service but once we start putting the transport layer in, whether it's layer two, layer three, you've got the network-to-network interfaces and then you're going into the cloud. How do you get those -- and we debated at lunchtime -- it's not really SLAs anymore, it goes beyond that. So those are the areas we're looking at and which we've discussed. Thank you.
Hi, thank you. Yes, David Howorth from Verizon. So I don't want to go over some of the stuff we went through earlier but obviously the issues that the two colleagues to the side have mentioned are obviously key considerations. I think I want to take the position as really Verizon is a cloud provider and also a provider of delivering application management. And I think it's at that stage where one is talking to customers and actually trying to understand what end user experience they want to have that we start getting into the conversations around service assurance, around governance.
And certainly I think the key thing is to -- and it was mentioned before -- define the parameters and define exactly what experience you want to have and then work backwards from there. And I think that to your point about what CIOs are concerned with if you were putting critical applications in there, it's can I really trust a partner to deliver that experience every single time because the consequences are potentially not more than just an outage it could be [brand].
And one of the things that we've been addressing as we go to customers and we're looking at offering them for example maybe a solution around a platform as a service is really offering them different levels and tiers of service, each with their own characteristics, whether it be from the amount of redundancy, the amount of security as well as maybe the amount of application acceleration that sits there. But giving the business choice. And I think it's very important for service providers to have these sorts of conversations upfront so that everybody is clear and has expectations and then has the governance around it to make sure that the commitments that they signed up, that people are able to be measured on them and held accountable.
Okay, that's good. And Amadou, your perspective?
Ahmedou Oul Ahmed Sidi
I'm Amadou. I'm marketing and sales director for Mauritel, and operator in Mauritania, West Africa. We all understand that we are not at the same level of developing (inaudible) like in Europe or America. We still in -- thinking about to go or not to go and we have some caution and some concerns. The first one is how the cloud technology can impact the value chain for an operator and also how it can impact the operating model for telecom operators.
We know that cloud technology can enhance the mobility because we can access anywhere anytime to the information. We know that it can also reduce the costs for the CapEx and OpEx. It can reduce the failure rate of the [infrastructure] because we integrate all. But we still have some concerns about security because as all information and services are integrated in the cloud is there a possibility to access for my information by others. That means are we protected against the fraud, cyber terrorism (inaudible).
The second concern is how the cloud can help operators to innovate in creating services and how we can move. We know that for moving from legacy infrastructure to cloud there is a cost. If we compare this cost with what we'll have after migrating to cloud is the cost -- the comparison is positive or not.
And the big question, is the quality of service with the cloud -- will have an enhanced quality of service almost for the critical services like money transactions, mobile money, all this. We're still at this stage and we wait for suppliers and developers to convince us to go further.
Okay, good point. And I guess essentially not every cloud does have a silver lining and so partly to perhaps David's points there's a need to see differentiation and different levels of service, some need more security and more reliability than others.
Can each of you give perhaps some comment into that as to how you think services might be discriminated going forward and how there might be some variation in that difference -- perhaps different -- I don't want to use service levels and service level agreements but different qualities of experience perhaps and different qualities of service. What sort of things do you think might happen?
Unidentified Panel Member
Well I do think it's going to be extremely important -- and as a matter of fact we're running a trial in this country -- to have some mechanism of content detection. But really understanding what type of content is basically hitting the network at one moment in time. So it could be real time video, real time video obviously is being very delay-sensitive and the delay variation is also very sensitive. Voice I think is obviously clear, any conversation that has more than a couple of milliseconds delay becomes unconversationable so you cannot move forward. On the other hand an e-mail, it could be a very important e-mail, but if it arrives 20 seconds later no one cares, right, no one knows when the e-mail was sent so it is not relevant when the e-mail arrives.
Mission critical applications that ask for a certain response time need to be detected and being advertised as such. So in the end it all starts with understanding which content is on the network and is accessing the cloud. And specifically like I was saying earlier in fixed environments it's a bit easier to separate because there's a fixed infrastructure. In a mobile environment you're all dependent on the capacity of the cell. And no -- and if you compare that to public roads no matter how much concrete you have on the public roads, at least in the country where I'm from, it's never enough.
And I give you one example is that last year the route I typically drive is one of the most congested routes in the country. They finally decided to add concrete to it, so extra lanes. And that went well for a couple of months and we're now -- after the holiday season we're back to square one. It's congested every single day. So the capacity that is available will be used and if you move into a cloud environment it is critical that that capacity, that access capacity is available.
So it starts with content detection and then with classification of the type of traffic that is there. So if you talk about SLA, if you talk about service assurance, the cloud provider, together with in the end the operator, need to be able to guarantee in service levels that a mission critical application can be used over any device at a certain moment in time. So in congested periods there needs to be priority being set for those applications.
And that is one of the things that if you hear a lot of the cloud discussions today it's all about okay, so can we move workloads from one server to another, are we flexible in terms of making sure that the capacity of a server is enough at any given moment in time. Security, right, very important topic and doesn't need to be underestimated but in the end it is about can you access that service at the moment you need it on the device because I think earlier this morning it was said that people are expecting that whether you've got an Android phone or whether you've got an iPad or whether you've got a Windows laptop or any device you have to be able to access your cloud services at that moment in time.
So it's critical to detect it, to classify it and then to ensure that you give priority to the right services.
And Tim, what's your view on prioritisation and so on?
I think you're right. We've talked about tiers of service that you're going to have. At the minute we tend to treat every bit of traffic which comes into the network,
particularly on the mobile networks, the same. There are ways we are now starting to look at that and how we can differentiate.
So for example a lot of our traffic going through is Youtube, people going in and accessing that video which we can store on (inaudible) devices pretty close to the base stations. So that takes away some of the workload which we can then use for the priority traffic coming through.
And I think we're going to have to be quite open about that. We're going to have to put classes of service in there for that traffic to actually be carried through the network.
I also think with the cloud providers they can't just turn around and say we deliver to here and that's the end of it. It is a complete end-to-end service and that's got to be provided to improve the customer experience. So again cloud providers, we're going to be asking them to see what they can provide in terms of -- again I come back to SLA but it's not really that, it's some form of guaranteed end-to-end service for the customer.
Okay, and so what will you be providing then David as a cloud provider?
Well it's an interesting -- just as I was listening to Tim there, I talked earlier about collaboration and to really -- this idea of being able to guarantee the end user experience as we all know, whether this be the last mile with copper or whether you're sitting in Tanzania trying to access your application over the local cell provider, that is a -- that's going to be a difficult thing. So obviously any service provider putting a metric on that is either going to need to have a lot of collaboration and obviously Verizon works closely with a lot of the -- we are in the US a wireless provider and obviously we collaborate a lot with Vodafone and other providers to deliver that overseas.
So I think that going back to your earlier point, from our perspective if we start off with understanding what is the end user experience that they want to achieve and where then obviously as a large network provider we would be looking quite often to utilise our own private network or the client's private network to actually backhaul that traffic to the most -- to the at least from a proximity point of view the best location to then potentially present that data out to the Internet.
And that's one of the solutions we look to do as well as obviously using CDM networks and then within the private network at least trying to prioritise some of that traffic. But it is a concern and it's something that I think that is not going to be just up to the service provider, it's also got to be a collaborative effort with the application developers as well in terms of developing applications that scale better and allow you to ultimately produce a better experience for the customer or the end user.
Okay. And Amadou, from your perspective?
Ahmedou Oul Ahmed Sidi
I don't have anything to add about that but I know that we, as I told you before, this is a stage of (inaudible) for services and we have to find the solution to answer this problem. That means voice (inaudible) have the same (inaudible) of data. In data you don't have the same (inaudible).
And another problem also, if we want to deliver services end-to-end on cloud that means we have also to review all the existing infrastructure because the capacity also has to be reviewed.
Yes, okay. And also as you mentioned earlier actually, the entire value chain. So there's so many people involved. We talk about cloud as if it was something, one single holistic item out there, but it's not of course. So many people have to collaborate and be involved together.
I'm guess -- I'm conscious there's only a few minutes, maybe five minutes. I'd like to open it up to questions if there are any questions out in the audience? Anybody awake after lunch? Alert after lunch? Inquisitive after lunch?
Unidentified Panel Member
They're trying to get on the Internet to access those applications.
Actually to paraphrase a UK comedy programme, the network says no. Sorry. Any questions anybody?
Unidentified Audience Member
(Inaudible) about latency. So as you spoke about assurance, have you planned to have different prices for different latency? Is it possible to manage that?
Unidentified Panel Member
I think for us the latency issue which we looked at, it was actually a realisation that by going into Asia Pac compared to Europe some of the applications we were hoping to run in the cloud wouldn't work. So we were looking at where we actually positioned our data centres to service those customers.
So for us the latency issue was that it was something we hadn't really considered before because we were in Europe and it was fairly close to all the locations. But when you start moving further afield into the bigger world, Africa and Asia, Asia Pac, then the distances start making a difference to the services you can provide and also the response that you can make for your customer experience.
So for example if you've got a data centre on the West Coast and you're running a service in Perth in Australia that currently wouldn't work on one of our services. So we've got to look at how we position those data centres. Maybe we change how the cloud works for that particular product and take that through.
I think the other thing on latency is yes, you could have a premium service which has very good latency and you could, I think as we said before, it's this tiering of services.
And latency would be one of those that you actually tiered at the high end. So again coming back to the question about e-mails, e-mails, if you get them five minutes or one minute it doesn't really matter. If you're a high frequency trader and you're trying to do financial deals, as we've seen with the Atlantic and between Chicago and New York, milliseconds count and people are prepared to pay that premium for that.
Unidentified Panel Member
But if you talk about different price levels I think latency is the mechanism but in the end you need to talk about applications. So what type of services can be charged at what price levels. Because different type of services -- again coming back to e-mail, e-mail could be free, so who really cares. You can access your e-mail, you can browse the web at any time but as soon as you need a guaranteed response time or you need very low latencies or you have a voice or a video service that is real time then obviously the tariffing needs to be different.
So instead of looking at latency, it's a measurement as a mechanism but in the end the user will understand which service has value to him and which not. And I always say a video can look beautiful, you've got a 1080P video on your iPad or 720P I believe, but anyway, so you've got it but when it stalls it's worthless. So it's not about the quality, it's about the flow of the application and that's how pricing can be set and should be set in my view.
Okay, any other questions? No, okay. A quick question from me then. And hopefully we only have a minute so a fairly short answer. Who's the hardest person to convince about the value of service assurance? Is it somebody inside your organisation? Is it somebody in one of your partners or is it somebody in the customers? Just a quick answer as to who you think it is. Who is it you've got to convince about this issue?
Unidentified Panel Member
I think if you're delivering a solution or a service that as we talked about is a collaboration of partners you can instil that customer-centric approach from your own organisation and getting people passionate but at the end of the day it's how you manage those complex supply chains of partners is the challenging thing.
Okay, that's good.
Unidentified Panel Member
Well it's really funny because we are providing services so value-added services to service providers over the cloud. So we deliver the core platform and then we deliver the service over the cloud. And the discussions we're having at quite an early stage right now, we launched our cloud services this Monday so this is quite timely, and the discussions we're having with our customers is quite funnily the hardest one to convince at this moment in time is marketing. And it sounds weird but it's marketing that's really worried about how can I run my service and how can I make sure that the customer satisfaction is there, that my customers are happy with that service.
Okay, interesting. Tim?
I think if you're talking internally about outsourcing your services within the company then it's probably the CIO. And I know from our point Albert Hitchcock who's our Group CIO would be a little bit concerned if he was the first one where it failed and it was a point we raised earlier. If they can see a success then all the CIOs will jump on the bandwagon because they can see the other things coming through. But I think it's that fear of failure maybe for being the first innovative mover.
Ahmedou Oul Ahmed Sidi
I think there's -- the hard person to convince to go is in charge of operating services.
Right. So again like the CIO. The person whose neck's on the block maybe, yes.
Alright. Thank you very much panel. Thank you very much audience and hope you found it a useful addition to the conference. Thanks.