Apps, Google Cloud and serverless: assessing cloud in 2019
- 09 January, 2019 03:00
Cloud computing is a fast moving beast, with new trends and technologies popping up all the time.
Last year we predicted serverless and Kubernetes to maintain their strong momentum, and that the 'big three' vendors would maintain their stranglehold on the market.
Some of that held true, but serverless continues to be much talked about but little deployed, and Microsoft and Google did make some inroads into Amazon Web Services' dominant market share over the course of the year.
This year we don't expect serverless or Kubernetes to go anywhere but they will continue to evolve as adoption ramps up and enterprises look for ways to leverage these new ways of working.
In a blog post for its 2019 cloud predictions report, Forrester analyst Dave Bartoletti has pegged 2019 as the year of widespread enterprise adoption of cloud to power digital transformation efforts. In the report, by Bartoletti and Lauren Nelson, the analysts go on to state: "In 2019, cloud computing will be shorthand for the best way to turn disruptive ideas into amazing software."
That's a lot of industry jargon, but it does have the research to back it up, predicting the global cloud computing market to exceed US$200 billion in 2019, which is up 20 per cent on last year.
As we take stock of the year gone by, here are some predictions for where cloud is heading in 2019, both in terms of the technology and the big vendors powering the industry.
Hybrid cloud momentum
With the release of Outposts in November, AWS finally admitted that it needs to be more hybrid cloud friendly for customers that will have applications hosted in their own data centres for some time to come.
It's a bit of a backward step for a vendor that has always been bullish (for obvious reasons) on the possibility of every app being ripe for cloud migration if needs be, but it seems like some customers have got their message through as AWS will now provide customers with a truly hybrid solution.
This brings AWS into better alignment with Azure, which has long been hybrid-friendly through Azure Stack, and Google claims to offer a bunch of tools to allow customers to stretch their applications out to the cloud, such as Kubernetes Engine and Compute Engine, as well as Stackdriver for holistic monitoring and Apigee for API management.
Add to that the major hybrid cloud play inherent in IBM's US$33 billion acquisition of Red Hat in October.
"IBM will become the world's number one hybrid cloud provider, offering companies the only open cloud solution that will unlock the full value of the cloud for their businesses," IBM CEO Ginni Rometty said in a statement at the time.
Stephen Line, VP EMEA at Cloudera sees that acquisition as part or a broader industry trend towards hybrid cloud and the need to provide customers with choice.
He predicts: "The hybrid model is a challenge for public cloud as well as private cloud-only vendors. To prepare, vendors are making acquisitions for this scenario, most recently the acquisition of Red Hat by IBM.
"Expect more acquisitions and mergers among vendors to broaden their product offerings for hybrid cloud deployments."
Hybrid is certainly a significant market segment, according to Forrester's cloud predictions. "Nearly 60 per cent of North American enterprises now rely on public cloud platforms, five times the percentage that did just five years ago," analyst Bartoletti said.
"Private clouds are growing fast, too, as companies not only move workloads to the top hyper-scale public clouds but create powerful on-premises cloud platforms in their own data centres, using much of the same open source software they can find in the public clouds.
"Success will be measured by developer satisfaction and time-to-market for new products and services, and not by taking out cost.
"Firms will build private clouds on top of what they have, build them on top of cheaper open source platforms, or have private clouds built and run for them. Whichever they choose, enterprises will get real about on-premises and hybrid clouds."
Gartner predicts that by 2025, 80 per cent of organisations will have migrated away from on-premise data centres towards colocation, hosting and the cloud.
Stephan Fabel, director of product management, Canonical predicts: "Despite considerable uptake already, we expect multi-cloud’s prominence to grow further still in 2019.
"Multi-cloud is almost becoming the default cloud strategy as organisations look to avoid vendor lock-in, granting themselves greater flexibility in deploying the most relevant cloud across different departments and functions."
Sean Fane, managing director, Spectrami UK believes: "Whilst the hybrid-cloud has been on the lips of CTOs for a couple of years, 2019 will be the year when we see organisations adopting true hybrid-cloud infrastructure, rather than multi-cloud environments.
"Concerns – whether justifiable or not – around data security and GDPR compliance will also drive adoption of hybrid environments – which not only give the chance to deliver the best performance at the best cost for each workload, but can ensure that data is always stored where it is most secure."
Google Cloud will make further inroads into the big three
It's very early days yet, but new Google Cloud CEO Thomas Kurian has a massive opportunity to earn Google a bigger piece of the global cloud computing market share pie, by making the vendor a better enterprise selling machine.
By investing in customer success and sales, Google will lose something of its traditional engineer-first reputation but will gain that all important trust of enterprise customers that need consistency and reliability in their cloud vendors.
As Ray Wang, founder and principal analyst at Constellation Research, told Computerworld UK at the time: "Enterprise customers need a different level of care, and Google hasn't been able to deliver to date.
"So the resources available to Diane [Greene] may not have always been allocated in the right place, but the resource is there and he has to sit down and see what partners and customer are saying."
If Kurian is able to better market Google as an enterprise vendor, rather than something of a machine learning specialist, you can expect the company to keep up its recent momentum and take more of the global market share from AWS and Microsoft and make things more of a three horse race.
In a more general overview of the market, Fabel from Canonical expects "to see Google focus on its AI credentials, Microsoft on its workload migration capabilities, and Amazon to continue pushing AWS hard in the public sector space".
The rise of the service mesh
With the release of AWS App Mesh and Google's open source Istio, 2018 saw the arrival of the 'service mesh'. We expect to see this technology gain popularity as more organisations look for a way to manage complexity and unify traffic flow management, access policy enforcement and telemetry data aggregation across microservices into a shared management console.
Owen Garrett, head of products at web server vendor NGINX says: "For organisations running large-scale, complex applications and who reach the limitations of extending their microservices applications, a service mesh promises a solution."
That being said there is still plenty of room for innovation, and uncertainty, around the technology.
"Time then will tell how it will develop as there is plenty of space for innovation," Garrett says. "Perhaps it will commoditise rapidly, and become a default, omnipresent feature of all major container runtime platforms.
"Perhaps new approaches, more efficient than the developing 'sidecar proxy' pattern, will emerge, offering better performance and lower resource usage. At this stage, there is no certainty as to how the technology will stabilise and who will be the leading providers."
And of application networks
The likes of Forrester have been talking up the idea of an integrated enterprise application network for years now, but the vendors themselves now seem to be buying into the idea.
"Salesforce, Workday, SAP, Oracle, and other app leaders are opening up their platforms and adding new development tooling, integrations, and deployment options, with an eye to which specific industries they can target to accelerate adoption. SaaS apps are becoming development platforms, and that shift will start to bear fruit in 2019," Bartoletti at Forrester writes, doubling down on a prediction he made last year.
He has good reason though, as Salesforce's record $6.5 billion acquisition of Mulesoft was a clear integration play.
Devang Sachdev, director of product marketing and solutions at SaaS vendor Twilio naturally agrees. "Application platforms will usher in an entirely new way for developers to consume and customise enterprise software," he tells Computerworld UK.
"Application platforms deploy like a SaaS application, integrate like an on-premises-based solution and iterate at the pace of API-based platforms.
"With application platforms, developers benefit from the low cost and scalability of the cloud are no longer limited like they are with SaaS which cannot be customised for specific business needs."
During AWS re:Invent in November we saw Amazon CTO Werner Vogels naturally focus on serverless computing, alongside a raft of granular announcements to help developers go serverless.
This included Firecracker, an open source virtual machine monitor for spinning up MicroVMs, Ruby support for Lambda and and AWS toolkit for popular integrated development environments (IDEs).
As technical evangelist at AWS Abby Fuller said: "I like seeing all the focus on developer quality of life improvement, so supporting the additional languages for Lambda, supporting Ruby outright but also custom runtimes to code in whatever language I want.
"It's really just being able to enable whatever use case developers want and letting them focus not on managing and provisioning infrastructure but what makes my application my application and what differentiates me."
We also saw our first (almost) full serverless deployment in the wild, with Danish web company Trustpilot speaking about their shift to serverless during re:Invent.
The latest Cloud Foundry Foundation global cloud trends survey found: "In just the past six months, awareness has swelled to the point that it is now in its next phase of adoption, with a plurality evaluating the technology." Across the EU specifically 17 per cent said they are already using serverless, with 31 per cent evaluating the technology.
What will be interesting in 2019 will be how this model, and the terminology 'serverless' itself, extends into the rest of the industry. Google and Microsoft Azure have cloud functions, so serverless deployments are a technical possibility, but neither vendor appears to have fully bought into the architecture yet.
Fabel from Canonical also expects "to see the beginnings of open source serverless solutions start to compete for broad acceptance in the developer community, which will shape the future of that technology ecosystem for years to come".
Ross Winser, senior research director at Gartner, for one predicts: "More than 20 per cent of global organisations will have deployed serverless computing technologies by 2020, which is an increase from less than five percent today."
Stephen Long, managing director for enterprise at KCOM predicts that "we will see much greater use of serverless design patterns in 2019".
"The major cloud providers all offer serverless runtimes," he said. "Of course, SaaS products are serverless to the consumer, but it is the ability to connect the SaaS and PaaS together without having to provision physical or virtual servers that is really the liberating idea."
Open source convergence
The back-end of 2018 showed a new trend towards enterprise giants going after open source communities.
IBM acquired Red Hat for US$33 billion and Microsoft bought GitHub for US$7.5 billion, despite confirmed interest from rival vendor Google Cloud. Not to mention the high-profile merger of open source data platforms Hortonworks and Cloudera.
So why the pricey acquisitions? And is this the signal towards a wider industry trend?
What it really marks is a drastic change of heart from previously renowned closed shops Microsoft and IBM that contributing to open source and developing proprietary intellectual property need not be conflicting interests, and that some of the most interesting developments are coming from that open source community.
The long term goal is to leverage these communities to stay ahead of the competition, but there is of course a short-term business benefit too.
As Docker CEO Steve Singh told Computerworld UK: "From building Concur, to being on the executive team of SAP, whenever you do an acquisition there's always a business reason behind it.
"The business reason might be I want great technology, the business reason might be I need to be able to monetise a particular thing.
"If you look at why SAP bought Concur, Ariba, or Fieldglass, or SuccessFactors or Qualtrics," the former SAP executive said, "they did it because they wanted to deliver that functionality to their customers, and they have this economic or sales engine or distribution engine that's just bigger."