Wednesday, December 18, 2013

Links for 12-18-2013

Posted by Gordon Haff at 3:16 PM 0 comments

Monday, December 16, 2013

Relearning past lessons about databases (& etc.)

There's an interesting interview with database guru Michael Stonebraker over at GigaOm. One of his points in particular caught my attention.
“My prediction is that NoSQL will come to mean not yet SQL,” he said.”… “Cassandra and Mongo have both announced what looks like, unless you squint, a high-level language that’s basically SQL.”
The perceived value of a purely low-level language all but gone, Stonebraker thinks NoSQL systems will also come to embrace ACID capabilities. It might already be happening.
“I think the biggest NoSQL proponent of non-ACID has been historically a guy named Jeff Dean at Google, who’s responsible for, essentially, most to all of their database offerings. And he recently … wrote a system called Spanner,” Stonebraker explained. “Spanner is a pure ACID system. So Google is moving to ACID and I think the NoSQL market will move away from eventual consistency and toward ACID.”
I suppose every new technology generation has to relearn the lessons of the prior one. I took a data science course earlier this year which, among other topics, spent some time going over NoSQL and "NewSQL" database technology. One of the clear trends was that a lot of the supposed baggage, such as ACID, that was ripped out of databases in service of performance and simplicity is now starting to get added back in many cases.
In Map-Reduce land, there are analogous trends. For example, Apache Pig "is a platform for analyzing large data sets that consists of a high-level language for expressing data analysis programs, coupled with infrastructure for evaluating these programs. The salient property of Pig programs is that their structure is amenable to substantial parallelization, which in turns enables them to handle very large data sets."
I guess it wasn't such a bad idea after all to build a lot of the optimization and parallelization in at the DBMS layer after all rather than forcing application programmers to handle it. On a side note, as someone who was following processor tech quite closely when multi-core hit the scene, I suspect that one of the reasons the "parallel programming problem" didn't develop into as big a problem as some thought it would be is that databases and other middleware (to use the term broadly) largely abstract parallelization.
I see this relearning of past lessons pervasively through cloud computing more broadly. Although, perhaps, reimagining is a better term. When we see the pervasive use of RESTful APIs, we're not really seeing SOA 2.0, although that makes a convenient shorthand. We are seeing a services-centric approach to delivering IT services. But it's a services-oriented approach that's much lighter weight and doesn't carry nearly the same amount of baggage as, say, a mid-nineties SOAP-based implementation. It's useful to understand why we did things or tried to do them in the past. It's also useful to understand why they might have been suboptimal or even ultimately failed--and why the environment (whether tech, ecosystem, or need) might be different now.
Posted by Gordon Haff at 12:57 PM 0 comments

Controlling Clouds: Beyond Safety updated presentation


I've given a version of this cloud computing governance overview at several events over the past year. I made a few tweaks and recorded about a 20 minute audio track to go with it (although you can get most of the content just by going through the slides.)
Posted by Gordon Haff at 12:03 PM 0 comments

Links for 03-16-2013

Posted by Gordon Haff at 9:08 AM 0 comments

Friday, December 13, 2013

AWS re:Invent: Using Red Hat's OpenShift PaaS to Develop Scalable Applications on AWS



Red Hat's Steven Citron-Pousty is a fantastic presenter. Here's the talk he gave on OpenShift at AWS re:Invent in November which rated #1 for the whole conference. Strongly recommended!
Posted by Gordon Haff at 2:59 PM 0 comments

Software is eating the world: software defined storage edition

"Is Your Storage Vendor Heading Towards the Elephant Graveyard?" was the provocative title of a presentation given by research directors Gene Ruth and Arun Chandrasekaran at the recent Gartner Data Center Conference in Las Vegas. I couldn't make it out to the show myself but one of my colleagues was sufficiently struck by this preso to share it with me.

So what struck us about the presentation?

Screen Shot 2013 12 13 at 12 18 55 PM

Certainly not everything discussed was unexpected. Flash memory has finally emerged as an important component of the enterprise storage market. Full stop. Gartner noted that it can even lower storage costs by allowing IT shops to avoid array upgrades and by lowering spindle count. (I'd argue that new technologies coming down the road such as persistent memory are also going to have a big impact but the bottom line is that we're finally moving beyond using essentially one class of storage media for everything besides backup/archive.)

What was less expected was the emphasis on open source and software defined storage (SDS). Now, don't get me wrong. It should hopefully be obvious that I don't need any convincing about the importance of these trends. I do after all work for Red Hat and we bought Gluster, the developers of GlusterFS, a couple of years back. GlusterFS (Red Hat Storage Server is the name of the commercial product) is an open source, high-scale, distributed filesystem that runs on volume x86 hardware. In other words, software defined storage.

But Gartner is often seen as taking a wait-and-see approach to disruptive new technology approaches. Its clients after all tend towards the mainstream to late adopters of technology as another Gartner analyst, Lydia Leong, told me a while back. Therefore, that Gartner is advising clients to put plans in place around storage trends such as SDS, open source, and hybrid clouds is noteworthy.

To be sure, Gartner makes it clear that the full promise of SDS isn't here today. That's fair. SDS is still relatively new even though Red Hat Storage Server is in production use at companies such as Intuit and Pandora. At the same time, though, Gartner doesn't equivocate about SDS's future. It recommends: "Begin evaluating SDS — it's a nascent concept today, but its time is coming."

Among the advantages of SDS that Gartner recognizes are:

  • Enables a vendor agnostic hardware infrastructure
  • Moves operations toward an SLA delivery model
  • Enables new hiring practices and skill profiles
  • Challenges conventional data placement thinking

Gartner also recommends that IT shops "insist that incumbent storage vendors develop and enable storage technologies that support hybrid cloud infrastructures."

Venture capitalist and co-founder of Netscape Marc Andreessen famously wrote about how software is eating the world. Software defined storage (like software defined networking) is just another aspect of this trend. The hardware doesn't go away of course. The bits have to sit somewhere. But the intelligence that places, replicates, and mediates the access to those bits is increasingly in open source software rather than in custom circuitry and silicon or the microcode of a proprietary vendor's array controller.

Posted by Gordon Haff at 12:18 PM 0 comments

Links for 12-13-2013

Posted by Gordon Haff at 10:59 AM 0 comments

Thursday, December 12, 2013

Links for 12-12-2013

Posted by Gordon Haff at 9:10 AM 0 comments

Thursday, December 05, 2013

Links for 12-05-2013

Posted by Gordon Haff at 3:24 PM 0 comments

Podcast: AWS re:Invent 2013 with Jane Circle

Red Hat's Jane Circle works with our certified cloud provider program and attended AWS re:Invent this year along with a number of other Red Hat folks. This year's event was twice as big as last year's and, among other things, featured an increased enteprise focus. Jane talks about what she heard from customers at the event and shares some observations about cloud security, cloud adoption, and cloud management.

Some relevant links:

Red Hat Storage Data Protection Workshops (tickets still available for Toronto and San Diego)
Amazon re:Invent

Listen to MP3 (0:12:33)
Listen to OGG (0:12:33)

Transcript:

Gordon Haff:  Hi, everyone. This is Gordon Haff, Cloud Evangelist with Red Hat. Today, I'm sitting here with Jane Circle, who heads Red Hat's Certified Cloud Provider Program. Welcome, Jane.
Jane Circle:  Thanks, Gordon. Glad to be here.
Gordon:  The reason we're having this podcast right now is Jane and a bunch of other Red Hat folks just got back from Amazon AWS re:Invent, out in not so lovely Las Vegas.
This is the second year of the show. Tell me, at a high level, what some of your impressions are and how things were different this year from last year.
Jane:  That's a great way to start. First of all, the attendance. We had about twice as many people, this year, at AWS.
They had about 8,500 people in attendance and about, I'd say, 200 sponsors this year, which was about a threefold increase from the sponsors from last year. It was a much bigger show.
From the AWS point of view, they really focused on introducing new services for enterprise customers. We also talked to a lot of our own enterprise customers in our booth and on the show floor.
Gordon:  I didn't make it out there this year, but I was listening to the keynotes over the Internet. I was struck, in Werner Vogels' keynote, how many of the things he announced and talked about were things that are very directly relevant to IT shops.
Jane:  Absolutely. Their focus is on commercial customers moving to the cloud, obviously, and still talking a lot about wholesale movement to the cloud.
What we found, when we talked to customers actually on the floor, is customers are still in the planning stages and just‑getting‑started stages, whether that's a one‑year or a three‑year project.
What they're most interested in, at least in talking to us at Red Hat about, is how they can manage AWS as a public cloud in conjunction with their data center resources. That was a big focus and concern for them.
Gordon:  Amazon even used the "hybrid" word up on stage this time, even though they still, of course, believe everything will be public and the on‑premise data center is a passing fad.
The amount of interest in hybrid out there is striking. It's even pushed Amazon to acknowledge that this is the reality if they want to sell to enterprise businesses.
Jane:  Absolutely. I agree with that. Certainly, our messages with open hybrid cloud resonate very well and worked in nicely with that talk, with Vogels' talk, as well as some of the other services that they're offering.
Gordon:  I was at re:Invent last year. Talking to people over by our booth, I would just say the majority of folks there were "Red Hat, you do Linux." Were things any different this year?
Jane:  They were hugely different. First of all, I didn't have any Microsoft aficionados coming by, saying, "What's Red Hat Enterprise Linux?"
There's a good understanding that Linux, and especially Red Hat Enterprise Linux, is the operating system to power the cloud. We didn't have to have that talk.
What we did find is that customers were very interested in cloud solutions. What can they take today and implement and then take into the future one year and three years out, and how can Red Hat help them plan for that?
When we talked to customers, we talked about our complete cloud stack, which was very important to them, that we don't just have point solutions. We don't have solutions that just, frankly, enable a service, a point service, or a point solution on AWS. We can help them with the breadth of our portfolio.
Gordon:  People often talk about developing applications for "the cloud," whatever they mean by that exactly. The popular image is of this long‑haired developer in sandals doing this DevOps‑y sort of thing, depending on what they mean by that word exactly. Is that a type of developer that you're talking to, or is it really a broader mix?
Jane:  It's a much broader mix. From our standpoint, we love talking to the DevOps guys. They came by, by the droves, to our booth and to talk to us.
We also got a fair number of administrators coming by, wanting to understand how to manage the environment. We had very good demos on CloudForms.
We just released CloudForms 3.0, which will give our customers on AWS much more manageability and flexibility in looking at that as a resource in conjunction with their data center resources as well. I would say it was a good mix between DevOps folks, as well as admins.
Gordon:  Speaking of CloudForms 3, in addition to supporting OpenStack, which we announced out at the OpenStack Summit in Hong Kong, one of the things that really struck me from the AWS support in CloudForms 3 was the support for virtual private clouds.
Which a lot of organizations seem to be really cottoning on to as though "Maybe this is the way we get that additional level of control in a public cloud."
Jane:  We've had customers using VPCs, for years now, on AWS. It works out quite well for them. We're actually working in proof of concepts with one of our large enterprise customers, who is moving much of their production‑level applications to AWS.
They're doing it all through VPCs. That added capability is definitely needed. Again, our support with OpenStack makes a big difference to our customers.
The fact that we're on the forefront of really understanding, whether it's through a proof of concept with our customers or production‑level applications, that we understand what it means to actually build applications on a cloud.
It's not necessarily cloud‑centric. What they're looking at is a global policy for security across their entire organization, and VPCs are one way to achieve that.
Gordon:  Probably being a bit rude to our listeners. We probably ought to explain, for anybody who doesn't already know, what a VPC is. Take it away.
Jane:  [laughs] A VPC, simply within Amazon, stands for virtual private cloud. It's a way that, in layman's terms, a customer or a user can wall off an application or an environment such that it's only accessible via firewall.
It's basically a straight pipe back to their own application environment. It really affords a level of security that, in a multi‑tenant, unmanaged cloud, doesn't exist in other clouds.
Really, enterprises need this level of security for them to feel comfortable moving to the cloud. Even when our customers are using VPCs... AWS will talk about this too... there's still this visceral feeling within security operations groups that it's still not safe.
That's not the case at all. VPCs offer a very high level of security for our customers. We feel comfortable with our customers using it.
Right now, we have actually moved a customer using our own security environment all the way to AWS. We have professional consultants that are very well‑versed in VPCs and how to use them, and so does AWS.
Gordon:  There certainly are some very legitimate reasons why folks will still want to run their own data centers. There are certainly governance issues, compliance issues, and various legal types of restrictions, on where data can be stored, for example. But the "Public clouds are insecure" really seems increasingly naive these days.
Jane:  I don't hear that much anymore. That may still, as I said, be a visceral reaction, but when you actually get down to the planning and the architecture, our security operations people readily embrace them.
We can do a reference architecture that includes VPCs, and our enterprise customers feel comfortable with that.
Gordon:  Let's move on to some other elements of the stack. Data is a big deal with the cloud. We talk about, for instance, having portable applications, portable workloads, that kind of thing, but portable data is, arguably, at least as important.
Jane:  AWS made some announcements around Hadoop at the conference. We actually just released a new version of Red Hat Storage, which supports Hadoop and Big Data applications.
We had the opportunity, at re:Invent, to hold a boot camp, which essentially was a six‑hour, full‑on training session for attendees that wanted to come and learn about how to move data and secure data.
Also, with these Big Data applications, scalability and flexibility is a huge issue. With the new version of Red Hat Storage, we've added features to allow for that, for these Big Data applications.
That was a great opportunity for us to connect with those attendees, hear their use cases, and work with them almost one on one during this training session. That was a really positive experience as part of re:Invent.
Gordon:  That's GlusterFS, for people who are maybe more familiar with the community name of the project. There's also a lot going on with OpenStack, with Gluster and Red Hat Storage Server as well.
Jane:  Absolutely. That's really exciting work that we're doing in the community.
Also, I should just mention that if folks want to know more about Red Hat Storage, they're actually running a roadshow now. I'm not sure all the cities that they're going to be in, but I'm sure we have information, Gordon, that you can provide to them on that.
Gordon:  There will be a link on my blog post, once this podcast is posted. I do have to give a shout‑out to one of our developer evangelists, who got the number one session ranking at re:Invent.
Jane:  Steve Citron‑Pousty. He is an amazing evangelist. If any of you have a chance to see him speak and to hear him, he's just fantastic. He's entertaining. He's energetic as all get‑out.
He conveys an enormous amount of information in a small amount of time. He had one of the main tent sessions at re:Invent. We were just thrilled when he got number one speaker. Some of the comments were fantastic.
He really took everyone through the paces of OpenShift, which is our Platform‑as‑a‑Service offering that comes in three different flavors. He really concentrated on the entire ecosystem and how to build a PaaS and integrate a PaaS on top of an IaaS stack like AWS. It was really fantastic.
Gordon:  It's pretty amazing to see how Amazon Web Services and re:Invent have come along. I was an industry analyst before joining Red Hat. I wrote my first research note about Amazon Web Services just in 2006, which is not all that long ago.
Now, you have a show in Vegas with 8,000 or however many attendees. It really is an important factor in the industry.
Jane:  It is, and it's international as well. Besides AWS having eight regions, the uptick in our customers wanting to understand how they can incorporate AWS and public cloud is across the board, in every region.
In fact, we had a contingent of our Japanese customers who came to AWS, which we thought was fantastic, along with a translator and everything. They were there to learn and to also understand where is AWS going to go for the next year.
We can clearly see that they're going to be adding more services, as they already have. We'll go right along with them.
Not to give short shrift, at all, to the fact that we have millions of customers running Red Hat Enterprise Linux, now on‑demand and also using our Red Hat Cloud Access, which is a bring‑your‑own‑subscription model, to move their subscriptions to AWS and build their applications there.
It's Japanese. It's over in Europe. It's certainly over here in the United States. Their fastest growing region still is in APAC.
Gordon:  Great. Anything else, Jane?

Jane:  No. It was a great experience. We are now getting ready for Red Hat Summit, which we can't wait for, coming up in April.
Posted by Gordon Haff at 3:12 PM 0 comments

Wednesday, December 04, 2013

Links for 12-04-2013

Posted by Gordon Haff at 10:03 AM 0 comments

Tuesday, December 03, 2013

Red Hat Enterprise Linux on Google Compute Engine

Google Compute Engine, essentially an AWS competitor, is out of beta.

Red Hat Enterprise Linux will be available on GCE--as it is on AWS and other clouds. This post by Red Hat's Jim Totton has the details on this collaboration.

Through this collaboration, we deliver:

  • Extended Choice: As a member of the Red Hat Certified Cloud Provider program, Google Compute Engine is a trusted and supported destination for developers, application owners, and administrators looking to benefit from the value of Red Hat. Additionally, as part of Red Hat’s open hybrid cloud ecosystem, Google Compute Engine provides the added choice of an Infrastructure-as-a-Service (IaaS) platform for customers looking to extend their environment in conjunction with a service provider.
  • Consistency: Red Hat Enterprise Linux instances on Google Compute Engine deliver the same features (e.g., performance and security), and lifecycle (release and update cycles) as on-premise environments. If an application needs the latest development and run-time packages provided by Red Hat Software Collections, or security provided by SELinux, Google Compute Engine provides the consistent features and capabilities for robust application deployments. 
  • Certification and Support: A core value of Red Hat offerings is the assurance of support and certification for the platforms on which they are deployed. This promise spans from hardware platforms to virtual and cloud environments, providing customers with support at all levels of the stack and across all deployment models. Red Hat and Google have worked together to ensure that the Kernel-based Virtual Machine (KVM) platform that Google Compute Engine is built upon, is powerful, capable, and supported when running application workloads on Red Hat Enterprise Linux as the guest operating system.
Posted by Gordon Haff at 9:35 AM 0 comments

Monday, December 02, 2013

The Amazon drone kerfuffle

I wasn't going to hop on as it's already all over the place, but a few quick thoughts so I can call it a day.

You may or may not buy the claim that this was orchestrated to distract from a recent unflattering book and other, albeit rather minor, PR hits. I'm not sure I do. What I do know was that this was pretty much a puff piece airing the night before the biggest online shopping day of the year. That makes it a huge PR coup in any case and one which 60 Minutes--which at least purports to be a serious journalistic institution should have taken no part in. For appearances sake if nothing else.

Amazon, like others, is seriously interested in same day delivery. As such, it only makes sense that they would be experimenting with and investing in various delivery channels and logistics approaches in support of that goal. In that vein, prototyping a delivery drone makes perfect sense. So would putting out feelers and even doing a little lobbying around the concept to feel out what's possible and what isn't--and maybe influencing regulations a bit even if, realistically, delivery drone-friendly regulation is years or decades away. Amazon has a history of taking the long view.

But, make no mistake about it, this is a long and speculative play--for innumerable reasons that many others have articulated. Fundamentally, I'm not sure when or if we as a society will happily countenance swarms of small drones (excuse me, UAVs) swarming through the skies. And it's hard to imagine them functioning in cities at all--in there words, precisely where population is densest.

Finally, we actually know how to do a pretty good job of same day, even 30 minute, delivery given sufficiently close "distribution centers." It's called Dominos. In all fairness, pizza shops can operate at a lot smaller scale than warehouses. But same day delivery doesn't have to mean 30 minute delivery and it's not hard to imagine some sort of multi-tier distribution system from a warehouse outside an urban core to local delivery people in the city or around a set of suburbs. To be sure:

Those with memories that stretch back a dozen or so years (or who have watched the documentary “E-Dreams”) will remember Kozmo.com. The start-up offered free delivery of videos, toiletries, and snack foods in about an hour, thanks to a fleet of drivers and bike messengers. The company raised $250 million — including $60 million from Amazon — but couldn’t make the economics of same-day delivery work. Kozmo was out of business by 2001.

Amazon is at a different scale point today and is probably much closer to making the model work--even without drones. Which brings us to the fundamental point. This isn't really about drones.

There's no problem getting stuff from point A to B--assuming the stuff is in stock at point A and assuming a route that can be traversed at some speed between the two points. A are called warehouses. The route is called a road. The "only" challenge is doing so in a way which is economically feasible--which is to say it doesn't lose the retailer money because the buyer is willing to pay the retailer's average costs of delivery, perhaps subsidized in part by the profit margin on the purchase.

That's one of the big lessons of the dot-com crash. It's great that people want things. I want things. But to get them, I need to pay for them. 

 

Posted by Gordon Haff at 6:04 PM 0 comments

Links for 12-02-2013

Posted by Gordon Haff at 5:00 PM 0 comments
Subscribe to: Posts (Atom)