Summarizing Oracle OpenWorld 2015 in a few headlines and key announcement is not easy. Because of the vast scope of the Oracle product portfolio (over 3000 products), the enormous variation in detail of the announcements (some very high over, some minutely detailed) and the contrast in tangibility, from generally available today (‘go and download from OTN’) to available perhaps at some unknown point in the fairly near future (but read this safe harbor statement first).
At the same time, the major themes and general direction were fairly easy to read. Keywords materialize – as they are repeated over and over again in sessions across the board. Some may be just the new flavors of the month, other will be long term additions to our dictionaries. Looking at those keywords may give a good indication of the real substance of the conference.
The next figure gives an overview – a fairly arbitrary one, as I did not use streaming analytics to intercept each and every word, aggregating in real time and visualizing the result in citizen data scientist’s dashboard. Even the highlighting in this figure is debatable. However, this should give you a pretty good indication of what the primary themes and hottest topics of interest were throughout the conference and across the stack.
The notion of multitenancy is spreading out. Initially this concept of supporting multiple mutually isolated consumers on the same platform service instance – to increase density (efficient use of resources) and ease the administration burden – only existed in the database (pluggable databases). WebLogic has now implemented multitenancy, with partitions, and SOA Suite and other Fusion Middleware components are soon following that example. Containerization can be seen as another form of multitenancy – allowing Linux Containers with Docker or Solaris Zones to coexist with fairly high degrees of isolation within the same instance of a compute unit (physical server or virtual machine).
As it turns out, the units of isolation prove to be very useful as vehicles of portability. Containers (as images), WebLogic partitions and Pluggable Databases can all be cloned, exported, distributed and imported into the same or another instance of the applicable host which can be on premises or in the cloud. This portability is a key element in enabling lift & shift [of workloads, to and from the cloud].
Many products and cloud services require a form of run time interaction across the gap between on premises and cloud. For example for integration purposes – GoldenGate, MCS, ICS and SOA CS – and for the collection of metrics and log files – Management CS – exchange of identities – IDCS – or to do run time administration from a single pane of glass [across environments on premises and in the cloud]. The keyword here is agent: five to seven different agents are being developed at Oracle Corporation, to enable some form of cloud on premises interaction. Note that in general these agents need to provide access to on premises systems to cloud services – and will do so to call out from the on premises agent to the cloud to get instructions on what to send. Even though it seems like internal systems are exposed to incoming requests – that is not actually the case. Similar to how (long) polling in web applications in the days before web sockets can feel like push.
The Oracle Cloud user experience is conquering ever more territory within Oracle. All cloud services and many on premises products have adopted elements from the Alta UI look & feel and the concepts from the simplified, mobile and extensible user experience. Oracle JET in addition to Oracle ADF plays a pivotal part in the actual implementation of the user experience.
The SaaS offerings and on premises enterprise application make extensive use of visualization, capturing on our human ability to absorb and interpret data and information much more rapidly – one figure beats a thousand words after all. An important goal with user experience is to achieve increased participation by making applications easy and even pleasant to engage with. Many enterprise applications see very low rates of user adoption (that is: the application being adopted by the users) and those that do use it, use it to a very limited extent and not seldom in the wrong way.
These applications also try to apply the glance, scan commit pattern, and leverage devices in the proper way. We scan our surroundings for things that stand out in some way and therefore may require more attention. The things that attracted attention on our first scan are then glanced at – to better understand what makes them stand out, quickly interpret if action is indeed required (such as more in depth investigations, decisions or execution of a workflow or business process). Some things can perhaps be done on the fly (accept/reject, send left or right, acknowledge) and are immediately removed from the to-do list. For others we may determine that nothing further needs to be done. Then there are more involved tasks that are the end result of the scan and glance stages. These tasks form a to do list of things that we need to work on – commit ourselves to – either instantaneously or at a later moment.
Visualizations not only cater for the glance phase, they can usually also play an important role in the scan phase. Visualizations can allow drill down, to analyze a little further and look in more details into specific areas of interest indicated during glance. Additionally, visualizations may go beyond just presenting the data and also support simple manipulation of data – for example adjusting a gauge or dragging elements on a time line or even a bubble chart.
Wearables like the watch seems a perfect fit for glancing – though not really for scanning and certainly not for commit. Smartphones are good for glancing and a fair amount of scanning too. With tablets we get a point where commit is within reach, although truly committed power users will probably use a big or multi-monitor desktop or laptop systems.
With users leveraging various devices from different locations and at different locations throughout the day, two things are very important. First of all, efficiency and user happiness are best served by a consistent user experience across all devices. The way information is presented, the way actions are performed – and of course the data and content themselves – have to be consistent across all devices. Second, actions started on one device at one point in time can be interrupted – loss of connection, meeting to attend, an incoming call of an important customer or of nature. At a later point in time and perhaps on a different device, this interrupted action should be available to continue and complete. Oracle introduced session portability as a mechanism, backed by the cloud, to capture and retain details on the current action and allow sessions to be resumed later on.
Working as a public cloud provider directly on the operations of its customers brings a new spirit into the Oracle product development organization. They start to feel almost a DevOps responsibility, for meeting SLAs around performance and uptime as well as resolving bugs. The ability to almost continuously improve the software and roll out new features, is invigorating. Instead of developing software that is to be released in about a year’s time and that will be adopted by far away customers in maybe two years’ time is nowhere near as inspiring as building a feature today that next month goes live on the public cloud – and can be tracked for usage as well. An agile development approach and organization of the development process according to Scrum is making serious inroads at Oracle. The new cadence of releases will be something like a bug fix train releasing every two weeks and a new feature train publishing every month.
Silicon is the foundation of all of IT. And to some extent, we return to those roots – with functionality directly embedded on the chip itself. To squeeze more capacity out of our systems, we resort to smarter chip architectures. By doing real-time decompression on the CPU itself, we can keep much more data in memory (in compressed format). By implementing aspects of SQL processing on the CPU itself, we can speed up operations like filtering and joining. And by embedding security – such as memory access protection and real-time encryption/decryption – in silicon, we can crank up security levels, achieve ‘always-on security’, hardly without degradation of system performance. The SPARC M7 is Oracle’s flagship chip in terms of speed and obvious metrics – on chip memory, number of cores (32) and process threads (256) – as well as in terms of functionality embedded in Silicon. The new SPARC M7 processor-based systems are the Oracle SuperCluster M7 engineered system and SPARC T7 and M7 servers.
One new kid on the OOW block this year was Apache Spark, a framework for cluster based computing. Spark can easily be applied to big data processing, much more easily than straightforward Hadoop Map-Reduce. Apache Spark can interface with HDFS (the Hadoop Distributed File System) and other data sources (Apache Cassandra and the OpenStack Swift and S3 APIs). Spark has capabilities for real time processing and for machine learning, making it an interesting component in the world of IoT (Internet of Things). A close associate of Apache Spark around real time message processing is Apache Kafka. Oracle mentions Apache Spark for example when talking about the IoT CS, StreamExplorer and the Cloud Adapters for ICS, SOA CS and SOA Suite as well as in the cloud services around Big Data.
Some controversy exists about the exact way to indicate the environment-that-is-not-in-the-cloud. It sounds like something with ‘premise’, but whether the writing should be on premises or on premise seems a matter of debate. The word premise according to the dictionary is reserved for use as a term in logic meaning something assumed or taken as given in making an argument. Premises is what is used to refer to a piece of [physical] property. Some avoid the issue by just using on prem as an indication. It seems that on premises is the actual proper wording, and that it will lose out against the far more popular on premise.
Announcements happen all over the place during the Oracle OpenWorld conference. No keynote is complete without a dozen announcements and other sessions by Oracle staff will contain their own announcements.
I have come to realize that an announcement can come in various shapes and forms, meaning very different things. It may even be a little like telling your friends and family you are having a baby: you can start telling people about it when the baby is nothing but a twinkle in your eye – and you not even have discussed it with your spouse. Or you make the announcement right after the first positive pregnancy test. Then again, you can wait with the news until people start to wonder about the mother to be’s circumference and no announcement is really needed at all. Then, when the child is finally born – has been released and therefore is available – you probably let the world at large know about it.
Announcements at conference can be like that ‘twinkle in the eye’ stage: it is a plan or even a vague notion, that has absolutely no status or substance, but we are testing the waters or trying to create a self-fulfilling prophecy. Or they can be far more serious – but still refer to something that may be true at some point in time, but then again may not really happen after all. A ringing announcement such as frequently heard from the biggest stages at the conference “today we announce….” is all too often followed by something that is really just a vision, far from an implementation that customers can start making use of.
A statement of availability or even immediate availability can also mean many different things. Usually such a statement does not clearly define where, to whom and for what purposes the availability applies. It may well be that the product or cloud service in question is available to internal staff only, or too a few selected customers, or with only limited functionality. Even general availability or GA – which is about as good as it gets – is to be close scrutinized. On premises software that can be downloaded from OTN is the apex of availability. However, cloud services that are GA, are frequently initially available only in selected data centers (US first) and therefore not globally available.
The bottom line is that any product announcement – even one that has the word availability in it – should be closely analyzed for what it really means. And keep in mind that most announcements are covered under the safe harbor statement – which means that they may never become reality.
Having said all that, most announcement are meaningful –either because they indicate a vision and longer term trend or of course they do reflect something that is real and relevant right now or very shortly. This next figure lists some of the most eye catching announcements.
On the left side of the figure are the cloud services – many of which have no or only very limited availability right. The movement and momentum are clear and I am pretty much convinced that we will see most of these services in the next six months or so.
The Private Cloud Machine for PaaS & IaaS is one of the potentially biggest announcements of them all: get the public cloud on your premises behind your firewall, with all benefits from pay per use and fully managed to rich and up to date functionality, always highest patch levels and supreme availability and scalability.
In terms of the on premises announcements shown on the right: many of these have already materialized – VirtualBox 5.0, MySQL 5.7, FMW 12.2.1, OpenStack V2, Oracle JET – or have been demonstrated at OOW 2015 in so much detail that these announcements leave little room for doubt.
Plotting the announcements on a roadmap
The next figure shows our attempt to plot the flurry of announcements on a timeline that visualizes the short term roadmap. The arrow and dash line indicate Oracle OpenWorld 2015. Every release to the left of that line was available before or during the conference. Everything to the right is a guess, based on rumors, slips of tongs, reads between the lines and sometimes more or less confirmed statements from Oracle staff.
When it comes to cloud services – as depicted in this next figure – the story is even harder to tell because of many nuances of available in the cloud. When a cloud service is available from one region or one data center, that does not mean the whole world can have access. Additionally, when a cloud service is available to one particular customer, it does not mean other customers can also get access – let alone trial subscriptions are available.
What this figure shows is an indication of roughly which services will be available when at least to some select customers in some area in the world. However, some services may fall of the wagon – temporarily or even permanently – and others may very well get added. It does not indicate general availability: apart from most SaaS offerings and probably the Java Cloud Service and the Database Cloud Service, very few services can really be said to be generally available at this time.
Oracle OpenWorld 2016 is announced for 18-22 September 2016. That seems a pretty firm statement, maybe the most reliable future date of them all.