Note: In classical antiquity, an oracle was a person or agency considered to provide wise and insightful counsel or prophetic predictions or precognition of the future, inspired by the gods.
Firstly, I would like to give John Troyer and Amy White a great big thank you for the invite to attend RBD2 at Oracle’s global HQ at Redwood Shores, CA. I’ve driven past Oracle’s offices on the 101 to/from SFO on a number of occasions and wondered what it was like up close. I want to especially thank Amy for her “herding of cats” by pulling a cabal of the industry’s top bloggers around the globe into one day. Initially, I was surprised to be included in those ranks. As you know I’ve been out-of-the-loop for a while. And it’s my hope this event marks an end to my hiatus and a return to my usual output and engagement. As you might guess it’s been quite journey. The best part is, its only just begun.
Secondly, I would like to thank the OCI – Oracle Cloud Infrastructure Team (Map OCI to AWS or GCP as equivalent acronyms) and Ravello Team (although I feel the distinction between the two will, as time goes by, become increasingly blurry). Bringing a larger group like this is not a small undertaking – and I see it as a measure of their commitment to the community that they are staging events like this – as well as giving all vExperts complementary access to their cloud – an offer I will be taking up shortly as a supplement to my homelab. My only regret was brain-fading in the late afternoon as we went thru two customer-case studies – by then I was rather battling the jet lag.
Right. Thank you’s done. Now to the meat and potatoes. I made copious notes during the event, and we were given access to a raft of PPTs as well as a roadmap which was under NDA. Only the roadmap is off limits. As it is I had a feeling that the Oracle folks were watching their words closely – which suggests there something brewing which is off-road map… I’ll dip in those later when I see something worthy of note – as often it’s the overall impression that stays clearer in the mind. Who knows maybe Oracle will buy VMware, but only once VMware has bought Microsoft after its merger with Redhat… :p
Anyway, I’m just gonna riff here and see where it takes me. And then when I’m done I will look at those PPT’s see if that inspires me more than just to copy and pasting slides into a blogpost for some visual relief…
Ravello isn’t just about running vSphere in a nested configuration on top of a cloud. So I’ve been out of the loop for a while, and wasn’t at RBD1. Before I took my gap year, Ravello had just been acquired. And the general consensus was – that just like a HP acquisition “That’s the end of them then”. Some acquisitions have habit of disappearing into the corporate matrix and never being seeing again. Yet, later when folks ask “why did we buy these people” – everyone who made decision have gone, and those left – just look blankly.
This isn’t, as far as I can tell the case with Ravello. I definitely see Oracle Cloud and Ravello seamlessly merging together as they each consume each other’s bits. Back in 2015, Ravello was an interesting hypervisor (plus some management bits) that you could run inside an AWS instance and in turn install VMware ESX, in some crazy N-levels of nested vInception that would cause the average sys-admins ears to bleed. There it was Fun. A handy way to run a vSphere lab in the cloud – without the need for a marriage limiting home lab. There were folks doing something similar like BareMetalCloud’s implementation of Alastair Cooke’s AutoLab. Since the acquisition Revello have pivoted their hypervisor to run on OCI.
Note: This slide shows the performance deployment options… So basically, what they are saying is Revello+OCI is ready for production, and Revello on AWS/GCP is really only good for testing and vSphere labs because the performance is sub-par based on having too many rings of vInception with underline hypervisor that’s to optimised. I don’t doubt these performance figures are achievable but I’m less sure about the “secret sauce”. If I moved my nested home lab on to bare-metal I could show improved performance stats too… It’s a kind of truism. Remove the roadblock things move quicker… There were lots and lots and lots of other improvements listed. But let’s faces it few vendor’s products get worse as the years go by. Features matter less than a good, stable product that gives good performance at a reasonable price. It grieves me to say this – but features don’t make or break a software purchase (within reason).
Oracle is serious about cloud. Okay, they aren’t building DC’s, that would be a silly waste of money. They are however, putting some serious spondoolicks in the game. The kind of money where you won’t see a return on your investment in just two years. Certain (ahem) vendors jump into public cloud and get out of the game pretty quickly when they don’t see traction inside of 2-3 years. I won’t mention any names. You know them. Just sayin’.
So where are they spending their dollars – in hardware and software development. What was noticeable was the hardware side of the house. I mean serious hardware – at the network level the sort of bandwidth where concepts such as QoS become irrelevant as advertising the spindle speed on a SSD – or the insulation properties of chocolate teapot.
Note: Some powerful specs.
Note: This is horrible slide – and has no place any deck worthy of a customer. Only marketing wonks who know diddley squat about IT create this sort of collateral. If a competitor or partner saw this – you have just kicked the ball towards you own goalpost as they proceed to demonstrate the performance and management capabilities. The delete key is your friend.
Note: Got to have slide that includes big arrow rising upwards. Sales people love this stuff!
Why is Oracle serious about cloud? You’d have to ask them to get the corporate line. I feel my wild conjecture is more fun to read, even if it turns out completely wrong. My feeling is Oracle has woken up and smelled the bacon. As an ISV whose revenue streams still come from product lines that were the companies “killer app” – they are witnessing growth in the public cloud (AWS, Azure and so on). It just so happens that these providers are competitors too – and once they have the customers workloads, the next logical step would be for Microsoft to push its products, and AWS to push there’s – despite the fact that both these vendors offer Oracle products-as-a-service. Does AWS/Azure exist to preserve Oracle’s bottom line? No. Would they take any opportunity to squeeze Oracle out of the equation? Why yes. That’s the nature of the ball game.
Most software vendors are built around a core-application that founded the company – if they are lucky they may stumble upon killer app No2, and No3 – but for the vast majority killer app No1 is the crown jewels. And believe me (God that sounds, like Mr Trump!) they will do anything to protect that revenue stream… [I’m pointing my index finger in the air. Kill me now please. It would be a mercy killing]
How is Oracle Cloud different from other Public Clouds? During the course of the day one of the delegates (I didn’t see who) said “So this is basically public cloud as designed by database people”. I found that a funny wise crack amusing not least because like all good jokes, there’s an element of truth. It goes back to the “serious about cloud” in the sense that the hardware specification is screamingly fast. So much so that big workloads should/could/would get a massive performance boost – in a “throwing hardware at the problem” kind of way. But more than that Oracle Cloud seems designed for the “Enterprise Types” who seem to be congenitally conservative, patrician, and risk-adverse. We all remember how hostile and sceptical DBAs were to virtualization. Well, guess what those DBAs are going to be same about any form of public cloud. That’s right – conservative, patrician and risk-adverse.
Note: Oracle supports Oracle products. Phew. There’s a relief. 😉
Lift “n” Shift and Move “n” Improve. There’s another philosophical difference between say Oracle and AWS. AWS is really a developers/DevOps cloud. AWS is about spinning countless cattle-instances, ideally interfacing with -as-a-services features such as S3, Lambda, into your data-pipeline. I was at the London AWS summit last year where every presentation ended with the exhortation to re-write your application to work with these native Amazon services – get to the serverless-model, and keep the number of EC2 instances you have to the minimum. There were lots of very big (truly global) enterprises singing the praises of AWS – saving money and improving performance. The only fly in the ointment – was for those businesses built-on shrink-wrapped enterprise software, where “the developers” were shown the door some time ago – the idea of re-writing core line of business apps probably fills them with horror, dread and fear – not least the open-ended commitment to an application re-write with unforeseen and unexpected snags.
Oracle’s approach mirrors to some degree VMware’s approach to virtualisation in the 2000’s. Back then we were P2V’ing existing apps unmodified into VMware ESX clusters OR else installing the apps natively, and just moving the data across – and later adopting “VM First Policies”. This strategy presented the least risk, and the least effort – and a “easy win” for the infrastructure team. So, Oracle’s strategy is to say to customers lift “n” shift those apps to Oracle Cloud which should in the move, improve them. My fellow delegates cringed at the “Move and Improve” corporate-marketing speak – but I rather like it. It sounds like being compelled to do more physical exercise more by a personal trainer!
Moving the needle. Will this approach work? Is Oracle going head-to-head with the 300-pound gorillas in the room – AWS and VMware. No, that would be that dumbest move since Harvey Weinstein slipped into a bathrobe. In this game you “pivot” away from large gorillas. Moving the needle is all about coming into the market – perhaps behind others who are more established with bigger customer basis in that particular sector – and getting noticed and at the very least a seat at the table when it comes to the discussion. So, what does any vendor (great or small) need to do to move the needle, and what are the consequences if you don’t?
- Have unique value prop – that no-one else has…
- Demonstrate massive performance improvements
- Demonstrate outstanding cost savings
- Ideally do 2 & 3 at the same time…
- Do 1,2 and 3 very quickly so you catch your competitors napping with their pants down! By the time you have secured a reasonable market share – you should have a sustainable profitable business which, if you diligently stick at it will sustain itself, or ideal grow at such a rate that the C-Class executives don’t get twitchy at year-end.
What happens if you don’t follow this plan. You have about 18-24 months to make headway before you customers move along to the next car-crash at the side of the road. Oracle are convinced they are doing No5 having added a truck load of features in 2017, and opening new regions this year…
Yes, I know like VMware, Oracle doesn’t have a great reputation for No3 (demonstrating costs savings). They are regarded sometimes wrongly as a “premium” product – a Lincoln, not a Ford. However, Oracle claim they are significantly more “cost effective” (you never say cheaper in this business) than AWS/Azure. The other thing I would praise them for is not grifting the customer for bandwidth usage – specifically on egress traffic.
I have a feeling one of the reasons “Lift & Shift” doesn’t really happen is because of the egregious costs of egress traffic. It’s a classic “Hotel California” play forcing customers to just take their data and rebuild their app back on premises or with a different cloud vendor. This is easy if you are DevOpsy with Containers, less so if you mainly accept the defaults and click the “next” button when building apps.
[Aside: Why have the terms ingress/egress suddenly become in vogue? In my day inbound and outbound just worked fine… rambles on until taken away by the folks in white coats the to the great IT rest home in the sky…]
Customer aren’t competition. During the day I asked who Oracle/Ravello thought their main competition was. I’m very pleased to report that there was no going negative and dumping on competitors. Temptation was laid in their way – and they did not take the bait(ing). Full marks. One classic line I heard when asking this on the vendorwag podcast was saying that their biggest competition was customers, and their loyalty to legacy approaches and legacy systems. Yes, that legacy SAN you bought last year. 😉
In fairness this classic evasion is to avoid awkward questions about competitors, you might be in co-operation talks with, and because the point is to talk about your bits, not someone else’s.
But there is another type of competition. It’s internal BUs, groups and individuals who see your project as a threat to their bonus or very existence. I’m talking about silo’d politics within large software/hardware vendors have grown into global players. One of the unspoken truths about software vendors – is whilst they cheerfully cajole and berate their customers about outmoded practises, intransigents silo’s and lack of “agility”. They themselves can be as riven with unseen infighting and politics.
Now, I’m not saying that this is the case at Oracle. I have no insight to that. But if there’s one thing that can upset the apple-cart is internal people who see a project as a threat to the existing shrink-wrapped on-premises business. There is only one way to overcome this as far as I can tell. Stuff people’s mouths with gold.
Salespeople with the best will in the world are “coin-operated”. If they have been at a company for a while – they get fat, lazy and complacent. They are so used in pulling the one-armed lever on a slot machine, and it coming up jackpot every time. They become unable to sell anything else. In the parlance of big corporates vendors – you need to “incentivise” your sale people, and “compensate” them richly to encourage selling ProductA to ProductB.
Is moving from on-premises VMware vSphere to Oracle Cloud Infrastructure (OCI) going to deliver the exact same feature set? No. But then again, no cloud provider offering that level of granularity from a feature perspective. Once of the points of public cloud is to de-complexify your environment where you focus just on the layer your responsible for.
HOWEVER, that does mean kissing goodbye to all the features, tweaks and tuning you have perfected and tooled over the last decade or more. I do hope customers are not going make the same mistake they made with Citrix, with VMware. But that is whole other blogpost….
Does OCI match pound for pound all the features you would get in AWS? No. There’s still plenty of work to be done. But they are moving quickly. The marketing people are going to busy with their arrows pointing ever upwards.
What barriers persist in convincing customers to look at the offering? There’s that persistent perception that Oracle “expensive”, but also an antagonism about how Oracle has reacted to the virtualization that happened underneath them. It’s always been a sore point in the VMware Community, and pretty much Oracle’s response has been – we don’t care – suck it up. In fairness that was no better or worse than Microsoft’s in the early days. But that is as they say ancient history.
And… well customers did suck it up – but only creating VMware clusters tightly packed with Oracle VMs to keep the “Oracle CPU Tax” down to a minimum. You see the trouble with unreasonable rules and regulations meant to shape human behaviour, is we creative learn to game them. But that same time get irritated by the unnecessary hoop-jumping and wanton technical compromises it generates. Oracle need to work hard to undo this poor perception. This is way above my pay grade – but maybe, just maybe – that’s the direction of travel. And if this is a road to Damascus conversion then should be welcome by all. In the world IT leopards should be allowed to change their spots.