The 5 stories you should read today: From CAP theorem to the DNC

After a very busy day, I'm going to keep it brief up top today and just link to five really interesti
ARCHITECHT
The 5 stories you should read today: From CAP theorem to the DNC
By ARCHITECHT • Issue #104
After a very busy day, I’m going to keep it brief up top today and just link to five really interesting items from today. They range from an assessment of the CAP theorem to a former Uber/Twitter exec joining the Democratic National Committee, so never say I don’t give readers a balanced informational meal ;-)
Democrats have hired Raffi Krikorian, a former Uber exec, as their chief technology officer (Recode): Raffi Krikorian was previously senior director of Uber’s Advanced Technology Center, and before that was VP of platform at Twitter. He knows a heck of a lot about building and operating cutting-edge web infrastructure at scale, and I’ll assume a thing or two about doing so securely, so it will be interesting to see how this plays out in his CTO role with the DNC.
The limits of the CAP theorem (Cockroach Labs): This is obviously a self-serving blog post, but it raises some good points that more companies are going to be considering in a world now populated by commercially available, globally distributed databases. Things get a little trickier to parse when you factor in that companies like Google own their own networks (which can improve availability) but that anybody can run a database like CockroachDB on the Google cloud.
Hedge funds love data and automation: OK, this is technically three separate items, but at least two of them stem from the same Future of Fintech conference panel. Long story, short: Hedge funds are investing heavily in data-savvy employees and machine learning, but still see a lot of benefit in human intuition:
No one really knows what AI will mean for the economy: What everyone seems to agree on is that AI will boost productivity and raise GDP for companies that can leverage it effectively. Who benefits from these increases is the question everyone is trying to determine, because a fast-widening inequality gap could be a very bad situation. Here are two stories addressing the issue:
What if the data science “skills gap” is just a hiring hot mess? (Fast Company): I’ll be honest, I would have guessed that companies had already given up searching for those unicorn data scientists who have every conceivable skill and qualification. They certainly exist, but they probably have very cushy gigs already—especially if they have some AI know-how, as well. The author here argues companies should focus a lot more on finding people who can apply data to business problems (still not a solved problem) than on finding people with the right résumé.

Sponsor: CircleCI
Sponsor: CircleCI
Sponsor: Linux Foundation
Sponsor: Linux Foundation
Artificial intelligence
I’ll say this about Salesforce: It did a very good job bringing developers onto its platform via Force.com and other efforts, and this could open it up to a lot of new intelligent integrations.
Sometimes I highlight things because I find them troubling. This is one of those times. It seems like a great way to automate away actually finding good talent.
The latest release of DeepBench includes metrics on inference as well as training performance. Expect this space to heat up as AMD fights harder for a spot at the table and more new architectures come online.
This is really cool research, via which DeepMind researchers observed, among other things, that deep neural networks tend to have a shape bias when classifying objects.
This doesn’t seem like groundbreaking research, but it’s an area in which we’ll see a lot of attention as companies try to generalize deep learning into broader applications.
arxiv.org  •  Share
This isn’t news to anyone, but deep learning and computer vision have a strong future in medicine. They have proven remarkably accurate at identifying various types of cancer and, now, at classifying the results of heart tests.
arxiv.org  •  Share
I’ll be honest: I haven’t yet listened to this interview with Murray Shanahan, but it sounds fascinating. I’m anxious to hear how someone immersed in real-world AI research tackles Hollywood AI.
Sponsor: Bonsai
Sponsor: Bonsai
Cloud and infrastructure
The cloud brokerage idea or arbitrage opportunity hasn’t been figured out yet, but Platform9 seems keen on giving it a try. Containerization will help solve some of this, but there still are folks who need to deal with the infrastructure level.
Backblaze CEO Gleb Budman spoke about this on the podcast a couple months ago, but here are some more details about how the $.005/GB service is being used. It’s not a replacement for S3 in a lot of cases, but it does appear to have some solid use cases.
This is a win for both companies, although I would argue Google brings more to the table than Nutanix over the long term. Kubernetes is the connective tissue between the platforms, as far as I can tell.
Speaking of clouds vs. on-prem … This survey is very biased in favor of Kubernetes, but the info around platform preferences is interesting. In particular, people prefer on-prem deployments to any given cloud provider.
www.cncf.io  •  Share
I don’t know a lot about Progress or Kinvey, but this is a $49 million deal to simplify the infrastructure part of building mobile apps. Apparently that was not a viable market, as Kinvey is not the first MBaaS player to get acquired.
This post is an interesting take on how cloud and cloud-native technologies are underpinning the shifts to AI and IoT. I would caution against betting too heavily against “last-generation” tech that’s only several years old though, as those communities have lots of money, users and reasons to remake themselves for the cloud.
This is probably a useful attempt at trying to simplify the process of managing distributed applications. I liked the historical reminder that Cloud Foundry was a startup acquired by SpringSource, which was then acquired by VMware, which actually launched the open source project we know today. All before Pivotal was a thing.
Sponsor: Cloudera
Sponsor: Cloudera
All things data
This seems like any number of other data integration/pipeline startups, including Segment. However, it’s still not a solved problem as sources and endpoints proliferate.
I don’t know about value of the WebSphere Application Server integration at this point, but I do know that Lightbend (formerly Typesafe) does very solid business on data-driven apps, and integrating with IBM’s data analytic and data science services at least seems like a good idea.
The tweet speaks for itself, but because I’m a lowly non-Gartner-subscriber, I can’t provide any more context. However, I do think there’s probably a pretty significant user base for analytics, generally speaking, that don’t require anything the scale of Hadoop.
twitter.com  •  Share
Sponsor: DigitalOcean
Sponsor: DigitalOcean
Did you enjoy this issue?
ARCHITECHT
The most interesting news, analysis, blog posts and research in cloud computing, artificial intelligence and software engineering. Delivered daily to your inbox. Curated by Derrick Harris. Check out the Architecht site at https://architecht.io
Carefully curated by ARCHITECHT with Revue. If you were forwarded this newsletter and you like it, you can subscribe here. If you don't want these updates anymore, please unsubscribe here.