Oh IBM, what are we going to do with you?

The week in cloud

The past few weeks were not great for IBM but they did not bring the bloodbath — 110,000 job cuts or about 26 percent of total headcount — predicted by one reporter.

As for the round of layoffs that did kick off, [company]IBM[/company] wasn’t officially forthcoming about the number. One insider who requested anonymity said a “few thousand” workers were affected, with costs covered by a previously announced $600 million restructuring charge. And, The [email protected]a union-affiliated advocacy group for IBM workers, put the tally at 5,000 as of late last week. In a January 26 research note,  Sanford Bernstein analyst Toni Sacconaghi estimated that the $600 million charge would cover a layoff of about 8,000 people. That’s not nothing, but it’s also not anywhere near 26 percent of IBM’s workforce.

The problem is no one thinks this will be the end of cuts at IBM, which has thus far managed to avoid the sweeping layoffs HP, for example, has endured over the past few years. And, as political analysts would say, the optics were bad — as long-time IBMers were getting the boot, CEO Ginni Rometty was getting a raise (and a bonus.) 

On the bright side, IBM last week said it won a big cloud computing deal with Marriott International. Details were scant but a spokesman said this is a three-year contract on which major rivals, including [company]Amazon[/company] Web Services (AWS), [company]Microsoft[/company] and [company]Oracle[/company] also bid. (I’ve asked those three companies for comment and will update this post as needed.) Update: Oracle and Microsoft declined to comment.

Here’s the thing about IBM: It competes with an array of competitors including old IT adversaries like HP, Oracle and Microsoft but more importantly it faces AWS, which has set its sights on the sorts of enterprise workloads that are IBM’s bread and butter. And AWS is not used to the sorts of enterprise margins once enjoyed by IBM (although I would note that people who think AWS is unprofitable are mistaken.)

But, what’s particularly concerning to some IBM watchers (raising hand here) is that the company is known for buying its way into new businesses, as it did with SoftLayer two years ago, then slowly sucking the new company into the overall IBM borg. Sometimes that works fine. But my feeling all along vis-a-vis SoftLayer was that IBM needed the smaller company’s nimbler startup mentality and non-IBM worldview almost as much as it needed its technology.

Here’s what I wrote when the acquisition was announced in June 2013.

IBM is a big, important company, but its ability to turn out innovative  stuff has been constrained by a hairball of legacy technologies. The question now is whether it will take what is good about SoftLayer and infuse that into the rest of the IBM cloud (one hopes!) or muddies what is great about SoftLayer. And, to IBM’s point, we are still early in the cloud adoption cycle and the stakes are huge.

So even though it’s normal for startup execs to fly the coop within a year or so of selling their baby to a big company, it is still concerning that a chunk of the SoftLayer brain trust has already left the building — co-founder and Chief Scientist Nathan Day left IBM/SoftLayer last April, for example. Former CTO Duke Skarda apparently left late last year although his LinkedIN profile is ambiguous. And co-founder and former CEO Lance Crosby left recently; Crosby was expected to stay until at least July, which will be the second anniversary of the acquisition closing. Those who hoped for a shake up to the IBM way wanted to see a SoftLayer person — or perhaps some other relative newcomer — lead the cloud charge. That is not going to happen.

VMware’s Bill Fathers: Businesses want a business-focused cloud

When it comes to enterprise workloads, another company IBM competes with is [company]VMware[/company]. On this week’s Structure Show, VMware’s cloud EVP and GM Bill Fathers didn’t pull any punches. In a world where many people equate cloud computing with AWS, Fathers is an unrepentant critic, saying that big companies are not convinced that they can run hybrid clouds in conjunction with AWS. He cites Harley Davidson as an example of a customer which tried to go with AWS for a new application and ended up coming to VMware.

Harley Davidson created a front-end iPad app for its CRM systems so dealers could check inventories. “They tried to do it on Amazon but physically could not connect it to their existing Oracle database of record from a networking perspective and they gave up,” Fathers said. “We used NSX [VMware network virtualization] to craft a connection from their on premises environment to vCloud Air and integrated it back to applications living on premises.”

Then the money quote aka fighting words:

“I am not spending a second working out how you solve what I think is an unsolvable problem of a client who’s marooned an application in AWS and is desperately trying to get it connected securely back to an on-premises app.”

(Send cards and letters to VMware please, but also feel free comment below.)

But there’s far more. Father’s guest segment starts at about 12:30 in.

[soundcloud url=”https://api.soundcloud.com/tracks/189580709″ params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]

Bill Fathers VMware Structure 2014


Hosts: Barb Darrow and Derrick Harris.

Download This Episode

Subscribe in iTunes

The Structure Show RSS Feed

Amazon continues to reach into your server room

Microsoft already has a huge presence in most companies’ server rooms. And it hopes to keep it that way or at least, persuade those companies running Windows Server, SQL Server et al. to — when the time comes to move to the cloud — opt for Microsoft Azure. Amazon, of course, has another plan for those workloads.

Last week, Amazon Web Services launched an update to a previously announced  System Center Virtual Machine Manager add-in  that will enable your admin, in layman’s terms, to vacuum those Windows workloads into the AWS public cloud. System Center is Microsoft’s management console for Windows environments, analogous to VMware’s vCenter. The original AWS add-in announced last fall, let admins manage both their on-prem Windows and their AWS EC2 instances “out there” from the same console.

I love how The Register characterized this sleight of hand:

“As of Wednesday, that plugin can import Windows virtual machines from on-premises bit barns into EC2. [company]Amazon[/company] reckons it takes just a few clicks and – POOF! – VMs disappear into the cloud.”

Remember, Amazon is also wooing [company]VMware[/company] admins with a portal that lets VMware vCenter users manage both in-house VMware workloads and AWS instances from one place that looks and feels familiar to them.

I think Amazon’s enterprise workload ambitions are even grander than it’s already signaled. Here’s betting that the Service Catalog  announced in November to ensure that only users authorized to access any AWS service will eventually extend to managing third-party application access on premises as well.

Conspiracy theory? Maybe. But that doesn’t make it wrong.

New managed VMware cloud from Rackspace

In other cloud news last week, [company]Rackspace[/company], one of the original OpenStack powers, announced a “Dedicated VMware vCloud” as one of its managed private cloud options.

The new [company]VMware[/company] menu option was described as:

a single-tenant, hosted environment that enables enterprises to take the next step in their virtualization journey by offering advanced automation, self-service, hosted catalogs and access to the vCloud API and vCloud web portal.”

This was reported as a sort of shocker since OpenStack originally launched — 5 years ago? — as a counterweight to VMware and AWS, but in reality, Rackspace has a long history of offering VMware-based infrastructure. And, on the flip side, VMware joined the OpenStack Foundation  in 2012 by virtue of its purchase of Nicira.

Oh, and don’t forget Rackspace also operates and manages [company]Microsoft[/company]-based private clouds. So this continues Rackspace’s attempt to distinguish its service and support of several core infrastructures as a differentiator.

Structure Show: Defending the data scientist

hilary masonHilary Mason was chief data scientist at Bit.ly, data scientist in residence at Accel Partners and is now CEO of research company Fast Forward Labs so it’s probably not a shocker that she thinks the title of “data scientist” remains valid. Here’s her take on why that is and what it takes to move big data concepts from theory to real-world application.

And, to hear more on these topics and others from Mason and a bunch of other data brainiacs, come to Gigaom’s Structure Data conference that takes place March 18-19 in New York.

[soundcloud url=”https://api.soundcloud.com/tracks/187259451?secret_token=s-4LM4Z” params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]



Hosts: Barb Darrow and Derrick Harris.

Download This Episode

Subscribe in iTunes

The Structure Show RSS Feed


GE tech guru high-fives the public cloud

Despite the success of Amazon Web Services, and the resources Microsoft and Google have poured into Azure and Google Cloud Platform respectively to compete with AWS, public cloud still has a bit of a perception problem. Many companies still view the use of shared, multi-tenant architectures askance — something not quite trusted for mission critical workloads.

That’s why it’s a fairly big deal when the CIO of [company]General Electric[/company], staple of the Fortune 10, says it’s “all-in” for public cloud. [company]GE[/company], after all is pretty much the reference account to end all reference accounts with its market cap north of $230 billion, and which offers everything from financial services to household appliances to cat scanners and jet engines.

In a Q&A in October, GE’s Chris Drumgoole told InfoWorld that “north of 90 percent” of all the applications deployed by the company last year were in a public cloud environment. Further, he said:

” We have a model where we’re operating outside of our four walls in someone else’s environment, but we’ve been able to ensure that GE data — compute, memory, and storage — remain single-tenant, even though we may be in a multitenanted data center.”

So there, public cloud haters. It was not all that clear how many public cloud vendors GE  uses but Drumgoole mentioned AWS and [company]Microsoft[/company] Azure in passing.

“…we really view ourselves to be a service provider to our businesses, so our businesses can buy from us or they can buy from others. The best way to think about it is if you’re my oil and gas division you can come to me, as corporate IT, and buy Amazon in order deploy your applications or you can go to [company]Amazon[/company] directly or you can go to Azure directly.”

But the vendor that got the biggest shout out was [company]Cloudability[/company], which monitors multiple clouds for customers and helps with cost tracking and assessment.

We’re big fans of a tool called Cloudability, which provides actual visibility into the data. If I want to see what something costs, I log into my Cloudability instance.

OpenStack installs 2015 regime

In other cloud news last week, on Friday the OpenStack Foundation announced 2015 board members elected from a full roster of candidates. Eight new individual directors — chosen from a list of 27 candidates — are Tim Bell of CERN; Russell Bryant of [company]Red Hat[/company]; Alex Freedland of [company]Mirantis[/company]; Rob Hirschfeld of RackN; Vishvananda Ishaya of [company]Nebula[/company]; Kavit Munshi of [company]Aptira[/company], Egle Sigler of [company]Rackspace[/company] ; and Monty Taylor of [company]HP[/company].

Members elected from the 8 companies at the foundations highest Platinum Member tier are: Alan Clark of [company]SUSE[/company]; Eileen Evans of HP; Toby Ford of AT&T; Van Lindberg of Rackspace; Mark McLoughlin of Red Hat; Todd Moore of [company]IBM[/company]; Imad Sousou of [company]Intel[/company] and John Zannos of [company]Canonical[/company]. Newly elected directors from the 17 Gold Member companies are Simon Anderson of [company]Dreamhost[/company]; Robert Esker of NetApp; Tristan Goode of Aptira; Steven Hallett of [company]Symantec[/company]; Chris Kemp of [company]Nebula[/company]; Boris Renski of Mirantis; Sean Roberts of [company]EMC[/company] and Lew Tucker of [company]Cisco[/company].

At a glance what is notable here is that Symantec, which just joined the foundation a month ago, is represented at the Gold level but Piston and [company]Yahoo[/company] are nowhere to be seen. Also, Randy Bias, co-founder of Cloudscaling is no longer on the board, although EMC — which bought Cloudscaling last year is represented by Sean Roberts.

One of three bylaws also approved by foundatio membership is a change to OpenStack trademark policy around the adoption of DefCore procedures, although from the verbiage it’s not clear to me what this means exactly, so stay tuned on that. DefCore is, for lack of a better word, a set of rules that defines what OpenStack is, what features it should include and certifications against which OpenSack implementations are judged. The goal is to ensure a degree of interoperability between different vendors’ OpenStack implementations — so one vendor cannot innovate changes that are incompatible with the OpenStack standard and still call it OpenStack.

Can policy be built into applications?

Wouldn’t it be nice if, when you enter your privacy settings on Facebook or other applications, you could be sure those settings will be enforced through the life of that product. So, when you don’t want your face tagged in photos, it cannot be tagged even by someone else.

That’s the type of problem MIT Ph.D candidate Jean Yang is working on. As things stand now, it’s very hard for developers to write privacy rules right into their code. Yang’s Jeeves project aims to make that less of a hairball going forward.Check out our chat with her about halfway through this week’s podcast. Oh, and she also weighs in on the Reddit AMA she and two other female Ph.D candidates hosted last month.

[soundcloud url=”https://api.soundcloud.com/tracks/186115767″ params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]

MIT Ph.D. candidate Jean Yang.

MIT Ph.D. candidate Jean Yang.

Download This Episode

Subscribe in iTunes

The Structure Show RSS Feed


Cheap cloud + open source = a great time for startups 

It’s all Docker containers and the cloud on the Structure Show

Mo’ money, mo’ data, mo’ cloud on the Structure Show

Why CoreOS went its own way on containers

More from Facebook on its new networking architecture 


Note: This story was updated at 10:02 a.m. PST to reflect that the Chris Drumgoole interview actually happened in October, not last week. Credit the reporter’s brain cramp for the mistake. His point still stands, but it’s not as fresh as I thought.

Verizon Cloud shut down passes 30-hour mark

As of about 6 a.m. EST Sunday morning, Verizon’s Cloud was on hour 30 of what could be a 48-hour planned shutdown to rollout a major service upgrade. The provider had hoped — and its PR team had expressed — that most similar closures last 24 hours or less.

One customer Kenneth White, a security architect who designs systems for non-profit and federal clients, followed through on plans to live tweet the outage.

[company]Verizon[/company] has said that the major upgrade being implemented now will make future upgrades less painful to customers. Updates on the work should post on this Verizon customer forum.

While maintenance and service upgrades are to be expected, and users appreciate getting advanced warning on such events, the prevailing sentiment is that a full day — let alone two days — of no service is beyond the pale.

There are several ways cloud providers can offer high-availability service including hot patching and live migration. It was unclear if these technologies are part of Verizon’s plan going forward.

Data center construction fire

In other cloud news last week, there was a fire Friday morning at the construction site of a new [company]Amazon[/company] Web Services data center under construction in Reston, Virginia. According to CNNMoney and other reports, Loudoun County firefighters were sent in at 10:12 a.m. and had the fire, reportedly started by welding equipment on the roof, under control in about an hour.

Mirantis makes its OpenStack plug-in friendly

OpenStack upstart [company]Mirantis[/company] last week rolled out its OpenStack based on the new Juno OpenStack release. Mirantis has also worked with Tesoro to certify its Trove database-as-a-service in Mirantis OpenStack 6.0. ONe of the key goals is to make it easy for third parties to develop plug-ins that will integrate easily with this release.

In October, Mirantis scored $100 million in Series B funding from Insight Venture Partners with August Capital, [company]Intel[/company], WestSummit Capital [company]Ericsson[/company] and [company]SAP[/company], bringing total funding to about $120 million.

Structure Show!

Steve Herrod, who helped [company]VMware[/company] become, well, VMware, is now at General Catalyst Partners where he’s looking for startups in proactive security technology, and other areas that will be key in next-gen enterprise infrastructure.

Given the current availability of capable open source software and cheap (near-free) cloud infrastructure, there’s never been a better time to be a startup, Herrod notes. But listen to the whole show. He knows whereof he speaks.

[soundcloud url=”https://api.soundcloud.com/tracks/184974081?secret_token=s-DR28d” params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]


Hosts: Barb Darrow and Derrick Harris.

Download This Episode

Subscribe in iTunes

The Structure Show RSS Feed


Structure 2012: Steve Herrod - CTO and SVP of R&D, VMware

Structure 2012: Steve Herrod – CTO and SVP of R&D, VMware

Amazon Web Services adds to its enterprise war chest

This week Amazon Web Services continued its enterprise-focused feature push with new resource groups and a tag editor for EC2 instances, which means that IT people at big companies (or even smaller organizations) will have an easier time isolating the resources they use.

The new [company]Amazon[/company] Tag Editor makes it easier for admins to group resources on a logical basis whereas before they had to tag them “service-by-service, region-by-region” according to the AWS blog post.

Admins can also then allocate a group of resources that share one tag or more and the group can span regions and services. This, said AWS, creates a “custom console that organizes and consolidates the information you need on a per-project basis.” There’s more here from IDG news.

Over the past six months or so, AWS has launched features to appeal to Windows admins, their counterparts in the VMware  world; directory services, and a service catalog for its cloud products that some think will soon encompass third-party services in the cloud and on premises as well.

New Google SDK for data heads

Also this week in cloud,  Google said it is open-sourcing the software development kit (SDK) for Cloud Dataflow, which itself was announced (in alpha form) at Google I/O in June. Dataflow’s goal is to make it easier for analysts, data scientists and developers to access large data sets for their work.

Per the [company]Google[/company] Cloud Platform blog post, the SDK

… introduces a unified model for batch and stream data processing. Our approach to temporal based aggregations provides a rich set of windowing primitives allowing the same computations to be used with batch or stream based data sources. We will continue to innovate on new programming primitives and welcome the community to participate in this process.

This SDK sort of compares to the AWS Kinesis Java client while Cloud Dataflow overall will contend with Amazon’s Kinesis service, which, as one developer pointed out “is available now, fully supported and not in alpha or beta.”

Google Cloud Platform logo

Docker boss responds to CoreOS fracas

Who knew that containers could stir so much passion? This week on the Structure Show, [company]Docker[/company] CEO Ben Golub talks about the dustup CoreOS started with its Rocket container news and touches on other hot topics as well. So check out Derrick Harris’ and Jonathan Vanian’s chat with Golub about halfway through, but if you want to hear about our take on how [company]Hortonworks[/company] and [company]New Relic[/company] were faring in their first week as publicly held companies, start at the beginning.

Ben Golub, CEO of Docker

Ben Golub, CEO of Docker

[soundcloud url=”https://api.soundcloud.com/tracks/182147043″ params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]


Download This Episode

Subscribe in iTunes

The Structure Show RSS Feed

The OpenStack opportunity: so how big is it?

Calculating how much revenue — or profit — commercial vendors wring out of open-source software is a tough nut to crack.

One reason is that the vendors themselves muddy the waters. There’s lots of handwaving on areas of investment when it suits their purposes but as to revenue? Well, um, never mind. (Even the cloud revenue numbers these companies put out are obfuscated, but that’s another story.)

It’s also hard because the revenue models for open-source software vary widely. Some companies charge for support; others offer both an open-source (i.e. “free”) version of the software under an open-source license and sell a commercially licensed version; some offer professional services; some combine one or more of these models. It’s complicated.

Seismic shift in sales model

One thing is clear; enterprise IT vendors accustomed to reaping fat margins on up-front sales of software licenses and then healthy margins on yearly support and maintenance re-ups, will have a very hard time replicating that profitability in the open-source world. For one thing, big customers moving to open-source are doing so at least in part to get away from that old model, which they deem as onerous, if not downright punitive. They don’t mind paying for stuff necessarily — in fact, they like the idea of an accountable vendor — but they don’t like being strong-armed into paying for stuff they don’t need or use.

On the other hand, it’s very hard for vendors to profit off of free software. For more on this check out a really interesting blog post by Dave Kellogg, CEO of Host Analytics, based on his analysis of the Hortonworks S-1 filing last month. In his view, the only company in the open-source realm to build a significant business on open-source software is [company]Red Hat[/company], and even Red Hat is itty-bitty compared to [company]IBM[/company], [company]Oracle[/company], et al.

But back to OpenStack specifically. How much money are the dozens (and dozens) of self-proclaimed OpenStack companies making? As a gage of participation, the OpenStack Foundation has 8 Platinum members; 16 Golds and 87 Corporate Sponsors for a total of 111 official stakeholders. But getting real numbers out of any of them on their revenue is a tough task.

Parsing out OpenStack revenue

Luckily, for this Week in Cloud report, 451 Research took a stab at it and last week issued estimates on the revenue picture for OpenStack. It’s call? OpenStack revenue will hit $3.3 billion in 2018, up from about $883 million this year — a 40 percent compound annual growth rate over those four years. It surveyed 60 OpenStack players and its figures include both public and private cloud. It said current revenue comes mostly from 30 of those 60 comapnies. Opinion differs but I would bet [company]Mirantis[/company] and [company]Rackspace[/company] are the two companies now making some dough off of OpenStack. The rest? Well, it’s still early days.
OpenStack Revenue Predition

Whatever you think of OpenStack, it does seem to have won the branding and marketing war when it comes to open-source cloud frameworks. That’s so although even some OpenStack proponents concede that CloudStack is a more mature solution.

“If you want cloud now, you could go with CloudStack, but these vendors are buying into the future, which appears to be OpenStack,” said Jay Lyman, research manager for cloud platforms at 451 Research.

If you take the $3.3 billion as gospel, and divide that by the 60 companies surveyed, that’s $55 million per company in 2018. Not chicken feed but not a blockbuster either. But most onlookers agree that the number of OpenStack vendors is in flux and that the consolidation we’ve already seen — Cisco buying Metacloud, EMC buying CloudScaling, etc. will continue. It just does not seem feasible that there is room for a hundred or more vendors to profit from this technology.

And, to be fair, skeptics find both the current and future revenue totals for OpenStack to be wildly optimistic. We’ll have to check back in a few years to see what happens and hope that by then the vendors will be more forthcoming about real numbers. One can hope, right?

Structure Show: Handicapping the clouds

Other than the OpenStack melee, most cloud watchers keep their eyes on [company]Amazon[/company] Web Services, [company]Microsoft[/company] Azure and [company]Google[/company] Cloud Platform.

If you want an informed opinion about the relative merits of all these players, who better to ask than the CEO of a multi-cloud management company? So we did. This week’s Structure Show guest was Sebastian Stadil, founder and CEO of [company]Scalr[/company] and he has some interesting thoughts not only on the big three but on other clouds from Oracle, [company]Joyent[/company] etc. Don’t miss it.

[soundcloud url=”https://api.soundcloud.com/tracks/180964822?secret_token=s-MTCRX” params=”color=ff5500&auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false” width=”100%” height=”166″ iframe=”true” /]



Hosts: Barbara Darrow and Derrick Harris

Download This Episode

Subscribe in iTunes

The Structure Show RSS Feed