Keynote presentation by David Robinson, Chief Technology Officer, ST Telemedia Connect and Jonathan Seckler, Director, Product Marketing, Dell Networking
Partnering for Success with OpenStack and Open Networking: Dell and ST Telemedia Connect
The second day of the NetEvents AsiaPacific Press & Analyst Summit, Singapore 2016, opened with a short presentation from Jonathan Seckler. He said the future of networking is not being built in the labs but in customer datacentres. What’s a switch, he asked, pointing out that today it consists of a server architecture built on industry-standard components and software. That is what open networking is all about, he said, adding that Dell now specialises in open networking.
Our OS10, announced in Jan 2016, is open and flexible, he said. He added that the base layer of the operating system was delivered to the open source community with the aim of creating a great user experience. Talking then about Dell’s partner ST Telemedia Connect (STT), he said the company has innovated to deliver services., and was joined on stage by David Robinson, of STT.
Robinson said STT is a Singapore-based cloud provider with the motto: One world, one datacentre. He said that STT delivers deliver five nines globally using a very small team, and that it embraces open source and open standards.
He defined open as not being locked into a vendor’s R&D schedule. We can innovate on or own and run OpenStack and Red Hat, deployed on Dell hardware. The experience was seamless despite one hardware failure, he said. STT chose Dell based on the hardware specs, and got it all up and running within a day. Robinson said STT did consider buying components from the cheapest supplier but it wouldn’t have been economic for us to build our own systems. Price and support were key.
For the future, open standards & open source is the way forward, he said.
Debate Session IV—The Next Generation of Open Networking and Open Cloud Computing: Is It OpenStack Everywhere?
Introduced and Chaired by Dustin Kehoe, Head of AP Telecoms Practice, Current Analysis
Panellists: Gint Atkinson, Vice President – Head, Technology Asia, Colt Technology Services Co., Ltd.; Jonathan Seckler, Director, Product Marketing, Dell Networking; Derrick Loi, Senior Director, DC Solution and Services, Orange Cloud for Business, Asia Pacific; Steven Davis, Senior Vice President, Global Data Centre, ST Telemedia Connect; Jon Vestal, Vice President, Product Architecture, Telstra
Dustin Kehoe opened with research showing that the technology industry is moving to open – open hardware, software, source, APIs, and interfaces.
The future will be a platform play, he said, and be about building modular datacentres. Two problems face us: networks which are still slow, and offer static workloads with manual configuration, and cloud, especially hybrid, which requires multiple layers of management. Solutions are still vendor-specific, vendors still don’t work together and want to lock in the customer. Even if you build your own IP, it leads to fragmentation.
As for SDN, IT process automation is the main use case. SDN is mature today, and we are seeing deployments in all areas of the globe including APAC. Research finds that the top three expectations are that it will improve network reliability, accelerate application deployment, and lead to better integration. Customers want open source to avoid lock-in but find it difficult to locate the right open source skills.
Jon Vestal, Telstra, said his company uses open source OpenStack with SDN. Growing pains included asking network engineers who knew proprietary systems such as IOS and having to teach them them Python! It was a comedy of errors, probably the worst thing we ever did, he said, but it had to be done.
Jonathan Seckler, Dell, said switches are servers from a hardware architectural standpoint. The trend is toward open networks, although back in 1990s, enterprise applications ran on mainframes – you wouldn’t do that now because it’s proprietary – but that’s still how it is in the network. Open source takes cost out of the maintenance and hardware acquisition.
Gint Atkinson, Colt, question whether such hardware was carrier grade. Installing new stacks etc takes decades – back in the day we did build routers using open source software. In the end you had to build it to look like a mainframe but it wasn’t as fast as a focused proprietary system. There are no open solutions to give under 50ms routing failover. SDN can fit in if you’ve got a minute or two but rerouting protection isn’t fast enough on open source.
Steven Davis, ST Telemedia Connect, said cloud grade services (eg Facebook) are all run on SDN. Another key issue is price, Cisco etc are too expensive.
Derrick Loi, Orange, said the key enterprise concern is digital transformation. So we want to keep the intelligence and provide end2end services. Also the network needs to be application ready and intelligent enough to automate, based on workloads. It’s about how we orchestrate infrastructure within the datacentre. We have integrated over 80 ope source applications and customers can integrate their own applications. Effectively it’s IaaS, all automated. So we link SDDC with SDN.
Q: Where are we at?
Davis said after the cloud, hyper-convergence is coming via a single portal to control all your clouds. We’re starting to see that coming now.
Vestal said carriers didn’t at one point meet customer needs but now they are offering services and products that allow interconnection.
Seckler said the model is outside the network industry. Enterprises needs partners to package OSS open source software to make it easier to manage and deliver a service. It’s not a DIY process for most enterprises, not realistic to do a Google or Facebook – it’s not about going to white boxes and recompiling every day.
Davis said it took the open source software market seven years to catch up with what Oracle was doing – PostgreSQL did it.
Atkinson said what’s needed is a network that supports a wide range of services.
Conference Debate V—Empowering SMEs with cloud services: What, How and Who?
Introduced and Chaired by Camille Mendler, Practice Leader, SoHo & SME Services, Ovum
Panellists: Bernie Trudel, Data Center CTO, Cisco Systems, Asia Pacific and Chairman of the Asia Cloud Computing Association; Chong Powmin, Group Head Enterprise Products and Services, MyRepublic; Kevin Pang, Cloud Director, Solutions Development, StarHub Ltd
Mendler said there’s an injustice in the technology industry – we’re looking at it the wrong way up – digital transformation improvement via tech is not just for large enterprises – the real opportunity is in transforming small enterprises. This is an opportunity and understanding of that is starting to emerge.
SMEs don’t have IT departments to make it all work. We’re going to be talking about what they need. But how to reach them? SMEs realise they need to change how they do business. And the economic health of countries depend on the SME – so they’re important.
Who can they trust? There’s no front runner so it could be a local tech reseller or the telco. The service provider needs to be that trusted partner? They’re not buying on self-serve basis – they need help. There’s a very very long tail of self employed or very small companies. There’s also a hollow in the graph of middle sized companies.
What sectors are they in? Retail, wholesale, manufacturing, professional & technical services, construction. The services they consume will be different and they require specialist knowledge from services providers.
Q: What services does an SME need to run its business digitally?
Kevin Pang, StarHub, said it depends on the business but the focus should be on manufacturing – though it depends on the local economy. The SP has a key role in terms of delivery of eg SaaS.
Chong Powmin, MyRepublic, said the government helps in this part of the world, eg Singapore. It has funds available to help the SME with cloud adoption. About 10k businesses have benefited – but it’s is a very small number out of 200k businesses in Singapore. Also, the tax department has a refund scheme for hardware eg network switches, and for services like document management. Take up is low – people are too busy worrying about their business! What they need is retail in a box – all the services they need including eg EPOS for $499 month. Entire business is then digitised – run by the service provider, who do it all, from fixing your printer etc.
Bernie Trudel, Cisco, said SMEs need to leapfrog from zero tech to get all their services from the cloud. But what about security?
Powmin said we have a programme that allows customers to sell our services as an agent to other businesses. Like the Avon lady model.
Pang said we go to very small resellers who have boots on the ground. At this level there’s a very small distinction between consumers and the business, it’s difficult to categorise them.
Trudel said it’s about what services they consume, they need security, wifi, broadband, analytics, face recognition… I like the idea of the bundle.
Debate Session VI—Making the Hookup: Pitching for the Best Way to Connect to the Cloud
Introduced and Chaired by Nikhil Batra, Research Manager – Telecom, IDC
Panellists: Matt Allcoat, Chief Architect, Asia Pacific, Middle East, Africa & Turkey, BT Global Services; Hisham Muhammad, Director, Global Solutions Architect, AP, Equinix; Chris Rezentes, Regional Manager, Partner & Product Strategy – Asia Pacific, Verizon
talked about cloud adoption in APAC – emerging market spending on cloud is 25% of those of mature markets, growth prospects much higher for emerging. Using cloud for digital transformation. Maturity by country – Oz NZ Singapore most mature, Thai Phil, HK Malaysia Indonesia least mature. Most benefits to come to IT operations.
Q: starting point to cloud journey?
Matt Allcoat, BT, said there’s lots of organisations – about 26% – at the startup, immature point. Everyone wants to unify cloud and bring it together. LOB managers go out and buy unconnected cloud services. People use ID management and cryptography to unify cloud services.
Chris Rezentes, Verizon, said you need to understand what the customer wants, and what cloud services deliver to the customer. People are still concerned about security.
Hisham Muhammed, Equinix, said as a cloud exchange, we realised that we’re at the forefront of where cloud providers congregate – they put nodes in our facility. The challenges for cloud exchanges is the advice that we give and the advice given by software vendors. Early adopters already consume multiple cloud services. This adds complexity. So it’s sometimes best for cloud providers to collaborate.
Allcoat talked about shadow IT: some company CEOs goes to their C-level execs and tells them to go out there and make it work – don’t worry too much abut rules. The CIO can’t just say no, it’s insecure, so we have to enable that and make it work.
Rezentes said that customers want end2end SLAs. We find we have to understand how they use the cloud, the customer is expecting network providers to know it all because we deliver the network, not a cloud service but if there’s a problem, customer asks us even if we don’t provide that service. We do have SLAs for our own services of course.
Allcoat said customers are asking for more, eg we can provide contact centre in the cloud, newsfeeds, Salesforce etc. They want to combine those services that work for their business. Not off the shelf items, but bespoke services – not infrastructure.
Muhammed said that there’s a need for a combined service delivery.
Q: Cloud service delivery challenges via SDN?
Allcoat said SDN is easy in a walled garden, the challenge is to do it in a carrier-to-carrier environment – need to develop new standards and agree who to cross charge which gets complex.
Rezentes agreed, saying that that level of network provider orchestration starts with vendors. Our strategy would be to include NFV and SDN services, and we need vendors who can use that same software for SDN to communicate between SPs.
Q: SPs have been investing to develop new services?
Muhammed said that we need to understand that the late-to-game SP faces a pricing challenge. It means you need get in at the strategic level. So whatever state you’re in, the perceptions of stakeholders is to understand what the expectations are across the organisation.
Rezentes said people will be using multiple clouds and this needs to be incorporated with the network. It will save a lot of headaches if you use a network provider to make it work.
Allcoat said it’s easy to take the professional services hit as part of the deal. People often focus on the dealmaker and ignore the rest of the company. Instead, you need to meet the right stakeholders and understand who in the company is right to talk to.
Conference Debate VII—From Millions to Billions of End Points: Stress-Testing the Cloud and the Internet of Things
Introduced & Chaired by: Anshul Gupta, Research Director, Gartner
Panellists: Naveen Bhat, Vice President & General Manager, Sales Asia/Pacific, Ixia; Derrick Loi, Senior Director, DC Solution and Services, Orange Cloud for Business, Asia Pacific; Ashwin Jaiswal, Head – IT Business Consulting & Practice (Telecom, Media & Entertainment), Reliance Communications; Amit Sinha Roy, Vice President, Marketing & Strategy, Marketing Centre of Excellence, Tata Communications
Opening the debate, Anshul Gupta used the analogy of raindrops which aggregate into floods of water, and which cities need to plan for. It’s the same with the IoT. Most sensor data is small but it aggregates. There will be 21b things by 2020, and data will double in volume every two years, so there’ll be 22ZB by 2020. We’ve seen CSP network outages such as EE in the UK and Verizon in the US, often due to human agency rather than technical failure – such as servers ad systems being overwhelmed with combined people uploading pictures from a football game. This is why we need testing. Not just for data volumes, but compatibility, functionality, and connectivity as well as performance.
Amit Sinha Roy, Tata, said testing is essential to provide a good user experience. LoRa helps with testing aspects of eg range, frequency operating parameters. Then that data comes onto our network and those of others. But that’s not the whole story. With a smart fridge, the cheese maker, fridge maker and others need to have standards to work together – but there’s a problem if a sensor misbehaves and spews out unnecessary data and hits the network, so there needs also to be a means for dealing with this.
Naveen Bhat, Ixia, said you need to think about how granular your device testing is, and about how they communicate. We provide capability for manufacturers to test using a simulated set of traffic from thousands or millions of devices. The network provider needs to know that this will work. Latency becomes an issue especially for devices that are life-critical.
Ashwin Jaiswal, Reliance, said yes, latency is a big issue, as are the number of protocols involved. We need backwards compatibility with multiple protocols because they have to work in multiple scenarios with multiple vendors’ systems, platforms and devices. Also, in manufacturing scenario, devices are deeply embedded inside other machines, so how do they communicate? Testing won’t be easy, and must be done on case by case basis. It’s all evolution right now. And no test can cover all scenarios, so there will be failures and we will learn from that.
Derrick Loi, Orange, said his company has tailored solutions to suit different verticals and use cases. One recent development is for big data: one hotel chain wanted to make sense of their customer profiles and how to manage by adjusting their resources to support them.
So we made our cloud analytics ready. The data needed to be processed before being aggregated but many branches of this hotel chain were ill-equipped to handle this level of data, so they needed to have their infrastructure upgraded. They can now run the IoT apps to pre-process the data. We made the hotel branches resilient to failure even if the network went down. We set up instances of Hadoop on demand. We set up backups, made the infrastructure application ready – in other words made it easy for users to set up IoT instances. They just had to turn it on and off – ie turnkey operation for a hotel branch IT admin. They use a catalogue from they can choose which application they want to deploy, and all this got us over several barriers to adoption.
Jaiswal said, with respect to to latency, we’ve all experienced this, which is critical for eg health devices. He gave an example of a path that was cleared to get a heart transported quickly from Kolkata to Mumbai by shutting down roads. Is this what is required?
Q: Need to ensure whether vendors are ready for testing?
Bhat said the confidence among CIOs for testing is quite low. But when security breaches happen – most likely due to inadequate testing – the estimate of losses when info is stolen is unknown for a long time. So enterprises need to take active steps to secure their networks, if necessary by regulations that force revelations of breaches.
Debate Session VIII— From SDN/NFV to Fibre Cut Protection, the Hottest Trends for Global Telco Providers
Introduced and Chaired by Jeremiah Caron, Senior Vice President – Analysis, Current Analysis
Panellists: Matt Allcoat, Chief Architect, Asia Pacific, Middle East, Africa & Turkey, BT Global Services; Gint Atkinson, Vice President – Head, Technology Asia, Colt Technology Services Co., Ltd.; Andy Solterbeck, Regional Director – APAC, Cylance; Stephen Tsang, Head of Managed Services and Enterprise Architects, Telstra; Helen Wong, Director, Partner & Product Strategy Asia Pacific, Verizon; Frank Wiener, VP Marketing, Wedge Networks
Jerry Caron asked: What are the challenges for telcos? We worry about them! But they are making money – he gave plenty of example of very profitable telcos – but what challenges do they face? Operators do understand the importance of SDN/NFV but there’s not much urgency. They want to grow the top line not just reduce costs. The biggest challenge is a lack of corporate focus – can upfront costs of SDN be justified? Also analytics – can they monetise all the data they collect? There are not many interesting moneymaking cases yet – so it’s early days. But from the cloud user’s perspective, the telco is not a priority. With respect to mobility, telcos are not focused much on this in consumer space, they are focused more on B2B. And no-one focused yet on 5G standards. With the IoT there are lots of opportunities – but where do you start?
Andy Solterbeck, Cylance, asked: can telcos truly innovate? They need to develop new business models but no-one knows what the look like. They need to be nimble when they’ve always been about long term investments and sweating them.
Matt Allcoat, BT, said we work with top F1 teams. We their run LANs, data processing, and data transmission to the back end in UK. It’s all about doing it for the customer from the front end.
Helen Wong, Verizon, said it’s about connecting to the customer at the business level.
Caron noted that some telcos such as are BT moving into content delivery.
Gint Atkinson, Colt, said we used to focus on niche markets such as the financial sector with high frequency trading with high performance storage and compute etc. It was very capital intensive. It also led us from low latency services into ultra-high capacity services for cloud providers. But telcos have massive investments into resources, eg dark fibre, they need to squeeze every bit of money out of those assets.
As example is gaming companies who need servers with hundreds of thousands of users who need low latency globally. So this needs a set of services providers who work together using, for example, redundant ETREE services. The connectivity provider needs to reach all the way into the cloud, which means working with cloud providers and their APIs.
Stephen Tsang, Telstra, asked about the impact of SDN. Many of our customers are mining companies – some of whom have lost $10b of business. This feeds back to suppliers. So we’ve pulled items such as SDN/NFV forward very quickly. Wrapping it into a business driver is the challenge. Cost savings are around operations and mining companies are becoming tech companies too, just like finance companies. So they value us in a different way.
Frank Wiener, Wedge Networks, asked who is responsible for security – the network provider? Network security is the user responsibility, you can’t assume the service provider will deliver security. He noted that adding a security service can add stickiness and revenue.
Solterbeck said security should come from the carrier as they have visibility into all the data. They can make money out of this – it surprises me they aren’t doing more of this.
Helen Wong, Verizon, said we are moving into virtual functions service for private clouds. We have cut down the number of vendors for VMS servers to two. WAN optimisation is our second priority after security – then threat monitoring and load balancing.
Matt Allcoat, BT, said there’s a great opportunity to provide OTT services such as security. We’ve also entered TV and mobile in the UK. Need to collaborate with other providers to make networks that give more customers what they want.
Wiener said there’s a logical extension to go from data to security – then offer it to others who aren’t going through your network.
Tsang said we know more about our customers than most utilities.
Guest speaker presentation by Grant Halloran, Chief Marketing Officer, Anaplan
Unicorn Vs The Giants
Unicorns are software companies worth >$1bn. We compete against SAP, Oracle and Microsoft. We are a cloud-based software company providing enterprise-level planning and forecasting. We help customers predict events, simulate decisions in response, then act. We can connect all that on a single platform and make it happen instantaneously. It compresses decision-making from months to days. For example, we help HP, with thousands of products and people, using 15bn sets of numbers.
How to optimise? It now takes HP 3 days to optimise where it used to take months. We enable them to issue quotas and compensation packages to salespeople globally.
Anaplan has 610 people, and $240m VC from Silicon Valley, and see three-digit growth every year. We have 80,000 high-value users.
We took on Oracle, SAP & IBM. We needed an incredible product not just a minimally viable one. Most VCs want startups to start in the US – we didn’t: we went global from the start. The giants created frustrated customers who saw how we could help them speed up their processes in a fundamental way.
That’s because enterprise software sucks. The vendors don’t think about the user experience or self service. The fee structure sucks too, along with IT dependency. It’s just bad software.
Many of these products were built to allow eg professional services companies to make millions out of customisation. So we wanted to change that and do something historical.
Ours is a patented technical modelling technology with a billion cells in one model – it runs in memory and calculates millions of data points in seconds. If we’d started only in Silicon Valley it wouldn’t have happened.
Halloran talked about founder Michael Gould who designed the technology. The system has low dependency as it’s easy to use, providing self-service. It’s about process and organisational change not IT. We’ve also built a community, a cult-like following for the brand. Globally, conversations are the same wherever we go, and lots of customers are global, so that’s why we soled problems globally. We have a passion to change the way that business works – the money is a by-product!
Keynote Interview & Audience Q&A with Grant Halloran and Jean-Baptiste Su, Tech Columnist, Forbes
There are only 140 unicorns in the world. We create simple ways to understand data. We enable foresight using predictive analytics, to help companies react faster – can now enable them to proact by understanding the future. We connect that data into a decision model to allow action to be taken.
SMBs: We go through partners to medium-sized companies, and partner with NetSuite. We need to package things more so have an online store for eg budgeting package. Smaller companies don’t really need our technology.
With that the final plenary session of the even ended, and lunch was called for.