By John Foley
At Oracle OpenWorld 2016, Thomas Kurian will outline the latest developments in Oracle Cloud.
Oracle is developing new capabilities in infrastructure as a service (IaaS), platform as a service (PaaS), and software as a service (SaaS), as it continues to build and expand its integrated stack of cloud services.
As it does, Oracle is making it easier for developers everywhere to access and take advantage of the full range of tools and technologies available in Oracle Cloud, including Oracle software as well as non-Oracle and open source software.
In advance of the Oracle OpenWorld 2016 and JavaOne conferences in San Francisco, September 18 to 22, we asked Thomas Kurian, president of product development at Oracle, to discuss the many new capabilities and innovations that will be of interest to Oracle customers, developers, and partners.
As Oracle ramps up its infrastructure-as-a-service offerings, what are the offerings for compute?
Thomas Kurian: Oracle's IaaS strategy is to allow customers, using software APIs and a graphical user interface, to define software-defined virtual data centers that run in Oracle Cloud. These software-defined data centers obviously include virtual networks to which customers can attach different kinds of compute and different kinds of storage. And we do four things: We give you very fast and very predictable performance and scalability; great reliability and availability; differentiated security; and a high degree of control and governance.
Compute is the building block, and we've got five flavors of compute services for different kinds of workloads. If you want the absolute top-notch performance and you want to install your own operating system, we offer a bare-metal service where you can come to an API and programmatically create a virtual network and attach to it high-performance and very dense servers.
Oracle President Thomas Kurian explains the range of compute options available with Oracle’s infrastructure-as-a-service offerings.
If you want elastic compute, you can go to an API and create a hypervisor-based elastic compute environment that runs Linux, Windows, Oracle Solaris, Ubuntu, or CentOS. If you want a SPARC environment, you can do the same with SPARC Solaris, where you can create a SPARC environment, create LDOMs and zones, in Oracle Cloud.
If you're concerned about isolation from a security or performance point of view—you want predictable performance and you don't want any contention with your resources from other customers—we offer elastic compute in a model called dedicated compute, which is an isolated rack of 500, 1,000, or 2,000 cores where you can get an isolated rack of servers but elastically configured through the same API.
And last but not least, if you've got VMware or KVM workloads and you want to migrate them, we have a solution called Ravello that allows you to migrate your existing VMware and KVM workloads to the cloud without changes to the virtual machines themselves or their network interrelationships. So we have a range of choices for optimal performance and easy workload migration from on-premises to the cloud.
What about storage and networking offerings?
Kurian: There are two aspects to networking. First, what does the access mechanism from your premises to our cloud look like? We offer three flavors.
You can configure virtual private compute, which means that your compute and storage services in Oracle Cloud are an extension of your company's IP network address space. You would see our cloud services essentially as an extension of your company's corporate network and its IP address space into the cloud. The benefit of that is, if you want to move a workload that is on-premises to the cloud, the migration process is seamless.
If you want to move a workload that is on-premises to the cloud, the migration process is seamless.
—Thomas Kurian, Oracle President of Product Development
Second, if you just need a virtual private network, you can use your Cisco, Juniper, Open/R, or a variety of standard VPN clients and connect over a secure VPN into Oracle Cloud.
If you need 10 Gb/sec or more in bandwidth to stream a lot of data into and out of our cloud, and you do not want that traffic to go to the public internet, we offer something called Fast Connect. This is the ability to have a 10-Gb/sec link into our cloud either over an MPLS circuit from your telecom provider or from our partner, Equinix, with something called Cloud Exchange.
Within the cloud, you can define virtual networks of various kinds. We've got a very efficient network architecture to allow you to have a core network that does not get oversubscribed and avoids contention between people, and also then to partition that securely where two different tenants are segregated logically so they do not have cross-talk between them.
We have five kinds of storage. If you're running an application that needs superlow latency, we've got nodes with very dense non-volatile memory/SSD or flash so you get superfast response time. In addition, we have nodes that have all-flash SSDs in addition to NVME for up to 6.4 terabytes per node. So you can run a big application, lots of data, in-memory, with very low latency from local disk, which can be either NVME or flash.
If you want to put in more than that volume of data, we've got high-performance block storage with spinning disks as well as with SSD or flash-resource pools. If you want to share data between your data center and our cloud or between two different data centers within our cloud, we have object storage. You access it through an S3-compatible API, Openstack SWIFT, or through our RESTful API. You can back up files, database objects, put it into the object store, then replicate it to different services in our cloud.
And finally, archival is useful if you want to store objects in the cloud because you want to save money and you don't need superlow latency recovery times for that data. You get extremely cheap storage. It's $12,000 a petabyte—a petabyte—a year. It's meant to provide you the best cost performance for archive in the industry.
These new services are being offered from Oracle's cloud data centers. Where is Oracle building data centers?
Kurian: Oracle has 22 data centers around the world. In North America, we have a western-region location near Phoenix, a central region in Chicago, and an eastern region near Ashburn, Virginia. We have a data center in Toronto and a disaster recovery in Calgary. In Latin America, we're in São Paulo, Brazil.
In Europe, we have two locations in the UK, expanding to three, and two locations on the continent, expanding to three. We have a location in the Middle East under development, five locations in Asia, two in Australia, and two in Japan.
We're opening a data center in China with a partner. We have a location in Singapore, and we're evaluating data centers in Korea, Turkey, and India. So we have broad coverage around the world.
What are these data centers like?
Kurian: In each region such as the US western region in Phoenix, we typically have three locations on three separate power grids, so if one power grid goes down, the other two locations keep running. We call that availability domains. Each availability domain is connected to the others over dark fiber so you get less than 10 milliseconds latency, and each region is connected with other regions over a fiber ring. So you have different degrees of redundancy and high availability that you can design within a region or across regions.
What is Oracle doing to enable open source and non-Oracle software stacks to run in IaaS?
Kurian: Our cloud does two important things. It runs any piece of software, whether that is Oracle, Microsoft, IBM, open source, or a packaged application. As long as the software runs on Linux, Windows, Oracle Solaris, Ubuntu, or CentOS, it can run in our cloud.
On open source, we're doing a lot of things with the community to get three kinds of things done with our cloud. One is to certify the most popular open source packages with our cloud. This could be Cassandra, Hadoop, Node.js, Tomcat, MySQL, a variety of NoSQL databases, Kafka, Mesos, Docker, and many others. So you can pick and choose your open source technology. There are about 480 open source packages we have certified on our cloud.
Second is we're working with a number of partners—for example, HashiCorp—who have test-and-development tools that people like to use as part of their continuous integration DevTest or DevOps cycle. So you could check in your source on GitHub, kick off a dependency-management view using Maven, run a continuous build using Jenkins. Once your build is complete, you can pack up your image using something like Packer, deploy that to our cloud, and instantiate the service. And throughout that cycle, we're working on automating the developer lifecycle across this environment.
The third thing we have done is a number of optimizations on some of these key technologies for performance, reliability, and ease of use. For example, we've introduced a service to allow people to use Hadoop, Spark, and Kafka analytics completely elastically, with great performance and via an API. So you can push data into our object store and spin up a Hadoop cluster. You can also pump in data into our cloud using, for example, a Kafka stream so that you can do real-time streaming of data. On top of that Hadoop cluster, you can spin up Spark SQL, Spark streaming, and Spark Machine Learning (Spark ML) applications.
So in broad brush, with open source we're doing three things: certifying the best open source products. And it's an ever-expanding universe so people have a fantastic experience with them. Second, we're optimizing a bunch of them for our cloud and building services based on open source. And in many cases, we contribute back to those communities so that they get a rich set of functionality. And third is we're working with the leading tools vendors in open source to optimize the DevOps cycle and workflow.
What differentiates Oracle's cloud infrastructure offerings from those of other cloud providers?
Kurian: Five things: First is unbelievably fast performance. Customers have benchmarked us against other cloud vendors and found we're three to nine times faster, due to a combination of hardware and software architecture, and an ultramodern core network design. Second is scale and reliability. In our cloud, you can run mission-critical application and database workloads with great reliability. For instance, we can run databases as large as 240 terabytes in our cloud.
Third is the issue of control. We have capability in our cloud around something called subaccounts and compartments. It's about one simple thing: We want the IT department in a company to be able to procure cloud services from Oracle, give them visibility so they can procure centrally, delegate budget and quota to different departments and projects, then track utilization without being worried that people are buying things outside of their control.
Fourth is deep security. There are a number of things we've done from a security point of view which are truly unique. Assume you have two customers, each connecting to our cloud services from their own virtual private network. We have a flat, highly scalable core network, but every customer's traffic is isolated inside their own logical virtual network. So there's deep security capability built in through the entire stack, whether it's around data security, certificate management, how we handle encryption, or segregation of duties.
Last is ease of migration. Many customers say: 'We'd like to go to a cloud, but we have many existing workloads we need to migrate. How do we do that?' A common example is VMware or KVM workloads. We have a technology that allows you to look at your entire VMware estate, pick a collection of virtual machines that are interconnected, lift the entire collection of Virtual Machines as what we call an application capsule—all those virtual machines, their network relationships, their IP addresses—and lift all of them to the cloud as one unit without any change to the application, the VMs, or the virtual network. It saves you cost and time.
What is Oracle's platform-as-a-service (PaaS) strategy? And what new things has Oracle introduced?
Kurian: Our PaaS strategy is about solving three things. One, giving customers a complete platform based on Oracle and open source technology that's standards-based for building cloud services and applications.
Two, it dramatically lowers the cost of ownership for people using the software, because we've taken all of the cost and labor associated with installing, configuring, patching, backing up, encrypting, and maintaining the Oracle software and eliminated it by automating all of this through software. So you simply use the software; you don't have to operate the software.
Oracle Cloud's PaaS offerings reduce cost and complexity by automating many tasks associated with on-premises systems.
And third, it's built on top of our infrastructure as a service (IaaS), so you inherit the benefits of the IaaS from a performance, scaling, reliability, and security point of view.
There are eight categories to Oracle's Platform as a Service: data management, application development, business intelligence, big data, mobility, content collaboration, integration tools, and identity. And at Oracle OpenWorld 2016, we're introducing the ninth, which is a comprehensive suite of monitoring and management services for applications and systems.
So it's a broad collection of services, and at the heart, it's about this very simple proposition: Imagine that you wanted to use our database, but you didn't have a DBA or a storage admin or a network admin or a sys admin. In the past you would not be able to use it, but now you can because we've automated everything associated. And you simply can use it through an API or through a console that's as simple as an ecommerce system.
Oracle is introducing Oracle Database 12c for the cloud. What is so significant about this Oracle Database release?
Oracle Database 12c on Oracle Cloud adds new features and capabilities, including industry-leading multitenant architecture and new affordable database options.
Kurian: Oracle Database 12c has three important new capabilities or features. First, it's unbelievably fast and scalable to build really scalable cloud applications. In version 12.1, we introduced a capability for in-memory databases, and there are a lot of enhancements in 12.2 that we're announcing at Oracle OpenWorld that significantly improve the feature-richness of in-memory databases. And you can get access to it right away in the cloud.
Second, we've introduced a technology called multitenant database. This is the ability to isolate individual tenants' data from each other without having to operate many different databases—one for each customer. So one physical database instance can have many logical partitions, called tenants, and it gives you a much denser and hence lower cost system to administer because you can have many more databases inside one physical database instance.
The important announcement we're making at Oracle OpenWorld 2016 is an extraordinarily attractively priced version of this database. It's as cheap as $175 a month. And you don't have to administer or manage the database; you simply use it. And you get an isolated, secure multitenant partition.
Number three, there's a lot of new capability in Version 12c around improvements in data management and scalability. We have introduced sharding as a native feature of Oracle Database to allow developers to use shards as a solution to build large scale, scale-out architectures.
So it's a big new release, and it's designed to help solve some of the problems customers had historically running these scaled databases in the cloud. It gives you fast performance. It can scale really well both for data warehouses and transaction processing. And in addition to what we've done for the big, high-end customer workloads, if you're a small customer who wants to start with a simple app, does not have a DBA but wants to use the power of Oracle, you can now do it for as little as $175 a month.
What is Oracle introducing for Hadoop, big data, and business analytics?
Kurian: For big data, at Oracle OpenWorld 2016 we're announcing an ultramodern analytics platform that's elastically configurable in Oracle Cloud. Imagine you want to do analytics on data coming from your ecommerce system: What are people buying online? What are they browsing? You want to merge their clickstream, and you want to look at real-time data coming off your internet feed.
Oracle's Thomas Kurian announces a new Oracle adaptive intelligence application suite that's "a major step forward for Oracle and how we look at data processing in the cloud."
You can say, I'd like to load batch data into the cloud through our object store. Then you can come to a console or an API and quickly and easily spin up a Hadoop cluster. That will load the data from the object store, and you can run a variety of workloads on top of it—Map Reduce, Spark SQL, ML, or Streaming workloads.
And you don't have to administer the Hadoop cluster. It's done for you automatically. The real-time feeds you can stream into our cloud over a streaming service, based on technology called Kafka, which is highly scalable. And you can blend data between the Hadoop system and Kafka using Spark.
You can also do machine learning and other things. On top of that, if you are a data scientist, you can use this entire technology stack and pay for it by the hour or the month. If you want to spin up the workload, run the processing, and shut it down, you can.
If you are a business analyst but want to use this infrastructure, we're introducing a complete next-generation analytic tool set which has three parts, beginning with a data-flow environment, where you can model a data-flow pipeline. The data-flow pipeline can also call into a data-quality tool built on top of this infrastructure that lets you do things like cleansing, data preparation, and graph analysis, so you can keep your data set clean. And last, data discovery, which is an environment where end users can collaborate to analyze the data set. It's a major step forward.
On top of this infrastructure, we're introducing services like our Internet of Things, or IoT, cloud service. We're talking about a new generation of smart manufacturing applications. We're adding capability around Oracle Data Cloud, where you can blend applications and data together, and there's a new data-driven or adaptive intelligence application suite. So it's a major step forward for Oracle and how we look at data processing in the cloud.
What about application development with containers?
Kurian: Containers are a technology we've got extraordinary interest in. Docker offers three important benefits. First, it gives you a vendor-neutral, portable way to build and deploy applications. So you can design your application, package it, and put it in a Docker registry, then spin up your workload as a microservice within a Docker environment—all in a vendor-neutral way. It allows you to fundamentally not be tied to a particular cloud vendor from the point of view of how you create and spin up containers.
We're offering four capabilities. We're introducing a container service in our cloud: the ability to go to a console or an API and spin up a Docker container elastically and define policies if you want to scale up or down.
Second, you can have a Docker registry that allows you to stage images and other things in this Docker environment, which is fully managed. It runs on top of something called Mesos Marathon, which provides the infrastructure to scale out the workload.
The third thing we're doing is creating a set of prepackaged services that anybody who's building a microservice on this container can use without having to stand up their own infrastructure. For example, there are a number of things people want when they're building microservices. They want an elastic search solution so they can put their logs and get them indexed. We're introducing a telemetry service to do that. They want something called consensus to get coherence across a cluster. We're introducing, based on etcD technology, a consensus solution.
They want to have a streaming infrastructure to stream data in. You have the ability to stream data in over Kafka. So there's a lot of different pieces where our goal is to allow a person who wants to build a microservice-based application to not have to worry about infrastructure. Simply say, 'I need X amount of resource to spin up,' and we'll spin up a container for them. They put their images in the Docker registry, and we promote it up into the container that they've spun up.
And then instead of having to stand up all these ancillary services themselves, we give them multitenant versions of these services that they just simply call through an API and use as part of their application.
How can customers of Oracle's or another company's SaaS applications use Oracle's PaaS?
Second is Oracle Integration Cloud Service. Assume you have our cloud SaaS for customer relationship management, and you want to connect it to Oracle E-Business Suite as an enterprise resource planning system that sits on your premises. You can use Oracle Integration Cloud Service to do API Management, provide secure messaging using its service bus, and define orchestration or business processes on top of it using our BPEL solution.
Third, imagine you're running HR in our cloud. You're an HR professional, and you want to blend some data from HR with information from payroll. And say payroll is with ADP or Paychex or some other payroll system. You don't have a DBA or a professional analyst, but you want to do some analysis. So we've introduced Oracle Visual Analyzer or data visualization. It allows you to extract data from our SaaS app or from an on-premises system. All you need to know is Microsoft Excel to do an extract. You upload the data using a spreadsheet, and then blend it. All you need is a browser.
In order to do analysis and build reports, you do not need to have a DBA or a database, or need to know SQL. You can do all that analysis self-service if you're a line-of-business professional. So three ways you can use our cloud if you're a SaaS customer: to extend it, integrate it, do analysis.
Oracle is working with the developer community in new ways and offering developers new tools and cloud services. What are some of these?
Kurian: First and foremost, we're working with the developer community to make our technology accessible to everybody who wants to use it. If you're a developer, in the past, if you wanted to use Oracle, you had to have a server. You needed a DBA to stand it up. You needed to have a fairly powerful system. You had to pay for a license fee.
Now with our cloud, you can access any software we have with a browser or an API core, and we've published our APIs into a catalog. We make all these services available. You can trial these services for a month or two without paying a single dollar for them. So we've broadened the reach of our software to everybody so they can access it and build applications with it.
Thomas Kurian outlines new resources, tools, and events aimed at making it easier for application developers to take advantage of the wide range of technologies available in Oracle Cloud.
And we know developers love open source tools and technologies. We talked already about everything we're doing to bring these technologies and integrate them with services in our cloud so you can use the best tools to automate how you build, deploy, manage software environments.
Third, we're doing some specific things for the new class of developer called the infrastructure developer, the person traditionally called DevOps. If you look at infrastructure as a service, what it's really done is take hardware and made it programmable through an API, and there's a new class of developer who's responsible for spinning up infrastructure. So there are a lot of new things we're doing for them.
Last, we're launching a major new developer program. We're introducing a new portal, called Developer.Oracle.com, with a lot more refreshed content, very topical, covering things that really matter to the developer. We're introducing a new developer road show and training for developers in new ways. We're introducing a program to university students, where they can learn about technologies by running those solutions in our cloud and simply using their student ID to get access to these technologies. We're introducing a lot of online training, including these massively online courses for developers who want to hear from our product engineers about how a particular piece of software was designed and why it was designed that way.
So lots of new things. Making our software accessible to them, integrating with open-source tools, and getting them information and access to this technology.
We hear a lot about DevOps. What is DevOps and what is Oracle doing with its cloud offerings to make the DevOps function viable in companies?
Kurian: What is DevOps? When you move to IaaS or SaaS or PaaS, what we've really done is taken infrastructure—servers, storage, network—and made it programmatically accessible through an API. There's a developer who is responsible for creating that environment, for standing up compute, defining virtual networks, attaching compute and storage to it, and managing the whole process. That person has traditionally been called an infrastructure developer, or DevOps, because he's development but also responsible for operating that environment.
We're making our infrastructure and platform services accessible to a well-defined, well-documented, standards-based REST API so infrastructure developers can get access to these functions—create virtual networks, attach compute and storage, spin up databases or app servers—all at the end of an API.
The second thing we've done is integrate these APIs with popular third-party tools that people want, in order to make this entire devops pipeline highly efficient, highly parallel, and highly productive.
You mentioned Oracle is helping to train college students and other developers on the newest technologies. What are the next steps in this program?
Kurian: We've launched a student program, both for universities and for independent students who are pursuing continuous education. So if you're a student, and let's say you want to learn about Java or want to learn about how to use SQL within Oracle Database. We're launching a program to allow a student to go to Oracle Cloud, register with a credit card, a student ID, then use the software at an extraordinarily economical price.
Second, if you're a university professor in computer science and teaching a class, and you want a lab environment where you can train students on a particular piece of technology—and historically you had to buy a computer and install it on your floor—we now offer a program that lots of universities are working with us on to take these solutions and give access to an entire class set where the students can collaborate on computer science programs and projects using our cloud.
What's next in terms of simpler, faster access to Oracle Cloud services?
Kurian: Simpler, faster access. You know, the biggest new thing we'll be introducing this fall is a much simpler experience for a customer who wants to buy Oracle Cloud services. There are two things we're going to do. One, we're introducing a mobile app where you can go to an Apple or Google Play store, download the app, create an account, browse a catalog of services, and procure the catalog or the service. You can go through four screens, and you can get a cloud service up and running.
Second, we're introducing a new ecommerce store dedicated to our cloud services—again, a super-simple experience. You create an account, register your credit card, browse the catalog of services and pick one, tell us the configuration you want, and buy.
Earlier this year, Oracle introduced Oracle Cloud at Customer, a way of bringing Oracle's PaaS and IaaS services inside the firewall. How does this work? And what other cloud services can be offered this way?
Kurian: There are many customers who would love to use the cloud but don't for three reasons. First, the government of their country precludes data from any company or organization in that country from going outside the country's boundaries. And it may be that we don't have a data center in that country.
Even if the country allows it, in certain regulated industries—whether the federal public sector, telecommunications, financial services—there are additional regulatory hurdles that preclude companies from putting data into a public cloud.
Third, a corporation's own executives may be risk-averse. They may be fine with a class of applications running in the cloud, but for certain kinds of applications they may be too sensitive yet to put data into a public cloud.
So Oracle Cloud at Customer is a very simple notion. You love our infrastructure as a service. You want elastic infrastructure as a service or platform as a service. Exact same software, same API, same price, managed for you on your premises, contracted as a metered or subscription service, not as a bunch of hardware that you buy. There's enormous interest from customers.
The next thing we're introducing with the Oracle Cloud at Customer program is our SaaS offering. On top of infrastructure and platform as a service, you can also run Oracle Fusion application software as a service with the Oracle Cloud at Customer program.