Well, it all seems such a long time ago since I was in the Bay Area at the beginning of April. And I have been meaning to blog about my experiences and insights at Cloud Field Day for weeks. Sadly, I became very ill on the way home from the event – I came down with bronchitis. If you’ve never had it – count yourself lucky. It took me weeks to recover. After getting over that – a perfect storm of events overtook me – both good and tragic.
Anyway, I’m feeling MUCH better now, and I’m finding some cycles to work thru that “to do list” we all perpetual have… This won’t be my only blog on Cloud Field Day, as I intend to write a blog about the start-up that came out of stealth when we were there – called Droplet Computing. The are worthy of a blog all to themselves…
So there were a number of more traditional shrinked-wrapped software vendors at the event – and without fail each one endeavoured to show how they were pivoting their traditional stack to the cloud. It makes perfect sense to do that considering it was a “Cloud Field Day”. The clue is rather in the title. I want to be kind and say that all the vendors were at least trying to do that. But commonly it did rather feel at times as if ToysRUS were reacting to the onslaught of Amazon. And its not without irony how – where some traditional retail operators are really struggling to deal with the competition that Amazon.com brings in the domestic space, some traditional software vendors are struggling to deal with the competition that a public cloud vendor like Amazon bring to the table. There is an element of “lets wait and see” if this change really is happening – only to find that once the change has happened – you are on the back-foot from the get-go. It’s a much more dynamic and competitive landscape which is moving very quickly, and some of the traditional software vendors aren’t as “agile” (hateful word!) as their senior management might want to believe – or tell their shareholders and investors.
Now, this blogpost isn’t your usual – lets bash the old ways, and slag the traditional vendors off because they are easy targets. I genuinely would like to see their cloudy efforts succeed. Because I believe in competition. And I believe that a market that’s sown up by tiny cartel of big players is not in the customer or anybodies interest. Competition is good for the Big Evil Corporates too – because it stops them becoming lazy and complacent. So I would like to give credit where credit is due. NetApp for instance has gone down the route of effectively creating a brand new unit to deal with the challenges they are facing. Essentially, putting all the storage goodness they have into Amazon AWS. This is native enterprise storage in the cloud, with all the features and functionality you used to enjoy on-prem.
[yes, I said on-prem. I’m bored already with the language police going on and on about on-prem(ises). Can we devote our mental energies to something that is actually IMPORTANT, for once?].
Note: This is the interesting Vertias presentation, and is worth a watch.
At the end of the week (when delegates are losing the will to live!). Vertias came into talk about what they were doing. Sadly, the first part of the session was a bit of a bore fest. Until a separate team came on to talk – and the Old Skool Vertias guys had left the room. Now, what they showed us looked very much like a start-ups “minimal viable product”, and I’ve always hated that term, especially when its deployed by huge, huge companies. Guys, you need to up your game (not Veritas specifically..). MVP is a term beloved of the Valley and the Start-up culture. It’s rightful place is there – in the start-up culture, where smaller companies need to spend a little, impress a lot, react quickly – and critically attract VC cash. But I’m afraid MVP has no place to play in a massive incumbent. People expect you to apply some spit and polish, and the expectations from customers are naturally higher. Meet them. Exceed them. Anyway, I digress.
There was some speculation amongst the delegates about why, Vertias had chosen the route they had. I made appoint that we were in Vertias CorpHQ, why don’t we, like, just ask them? Mercifully, we were spared the usual corporate flim-flam. The guy told use quite straight down the line – that they just could not get the features and coding built – quick enough from the existing team. So they built their own.
There’s critical message for all companies like NetApp and Veritas. There’s a right way to do this and a wrong way. Trying to build the new company – from the people and process that built the old, is by its nature like rolling a stone up a hill. You nearly always need to build the new company by incubating a new BU from within. This is not just some organisational chart activity – critically that group needs to be championed, and yes “protected” from other BU/SILOs, that would quite cheerfully strangle it at birth. So a bigly beautiful wall needs to be built around the company-within-in-the-company – and then it need stuffing with cash. The other thing that needs to happen is the sales folks who earned their commission form selling the “Old stuff on the truck” need to be “compensated” and “incentivised” to sell the “New stuff on the truck”. And if they can’t do that – you need to get new people who will ,because they have no stake in the previous model. As a friend quoted to me:
“Change the people, or change the people”
Think about that for a while…….
Anyway, I name check NetApp and Veritas as companies who I think may have woken up and – smelt the coffee and bacon (an odd mix, but actually can be quite tasty, although as a Brit I prefer a cup tea with my bacon butty). I think I would include Oracle Infrastructure Cloud (OIC) in that mix too. Whether NetApp, Veritas or Oracle – are able to overcome the baggage that comes with their brand is anyone’s guess. But I do think they are at least trying, and if they put enough weight behind their respective projects – there is at least a chance they will be successful. And it feels churlish not at least encourage and recognise effort, where effort is being made. Oddly enough OIC didn’t go down as well at CFD as they did at the Ravello Bloggers Day (same venue, similar people, almost the same presenters). I’m not sure quite why that happened – as there’s lots of positive things to say about their offering. Perhaps that’s because they led with Ravello (for which there’s a lot of love for in the vCommunity?) Oracle comes with a lot of baggage with the vCommunity – so there’s “perception” issue to overcome. Sadly, the Ravello Bloggers Day wasn’t recorded, and I think embedding the CFD video might actually do them a disservice. So recommend checking out my blog on that event:
I’m afraid I could not say the same about Riverbed. Sadly, their presentation lent too heavily on older sales plays, that had merely been re-jigged for a cloud era. You could tell they weren’t really connecting with their audience by the rigor mortis that was settling into the group. That also kind of came across with the lack of vim and vigour in their presenters. I hate to say things like that. Because it feels like a personal criticism, which I normally shy away from.
Things did brighten up with a presentation from one of their team members at the end of the session (Vivek Ganti) – who at least came across as someone, who felt a passion for the work he and his team were doing. This guys didn’t feel like he was just “going through the motions”.
Sadly, however, what he was showing was some of the automation that Riverbed have put into deploying their appliances into an Amazon VPC. In fairness if you were doing this all by hand using Amazon’s frankly horrid web UI you would have piece of work on your hands. However, if your doing public cloud “right” you should be using your favourite toolset of utilities to leverage the API. That’s what public cloud is all about. Not Old Skool SysAdmin (like me!) clicking and filling in dialog boxes, but a new style DevOps SysAdmins who can stand-up and tear down infrastructure with the flick of script. One of the goals of this DevOps Public Cloud is to accept that nothing is really ever persistent in the old style way – but should be by its nature volatile – so it is so robustly automated it can be destroyed and created in an isn’t. My point here is that this kind of automation is likely to make the average DevOps SysAdmin go “meh”.
Also, for me this issue is also a more important one – the value in any software – whether is on-prem OR in Da Cloud isn’t how easily it can deployed and setup. That, really should be a “given”. The real value is what that software allows the customer to do – which they couldn’t do before. Now, I guess you could say – standing up a multi-tier load-balanced layer that offers redundancy, and inspection of packets to ensure a smooth network experience can be difficult. I actually think Riverbed have a fantastic suite of products (although folks tell me they can be quite pricey). But I wasn’t really convinced by their cloud play.
Why was that? After all they aren’t really doing that much differently than say what NetApp were doing. But there was something about the vibe. Whereas it felt that Riverbed were sprinkling cloudiness over an existing product range. I got the impression that NetApp had made a genuine attempt to Cloudify their existing product ranges, whilst at the same time acquiring and investing in something net-new. I don’t want to label Riverbed’s approach as “Cloud Washing”. I guess the difference is in the approach. Whereas it appears as if NetApp wanted to deliver NetApp-as-a-Service (NaaS!) with Riverbed it was more like Virtualization 1.0.
Hey, lets put our existing stack into a Linux instance and spin that up in EC2/VPC.
I suspect what customers WANT (remember them? customers?) is all the features and functionality they used to enjoy on-prem, without any of the complexity of configuration, management, or having to deal with it all becoming rusty after 3-5 years, and having to lash out more cash to upgrade and forklift their unique bits. Shrink-wrapped software, without being tied up in cling film if you wish….
Anyway, I said when I started this would be just one post, with another on Droplet Computing. But I feel like banging on about the companies who are getting this right from the get-go. And that feels like another post entirely (although it is related…)