Blog‎ > ‎

Python, Cloud, Neural Networks, Deep Learning, 2FA, JSON API, Startups, Mobile Web, Kafka, Samza

posted Mar 29, 2015, 12:56 AM by Sami Lehtinen   [ updated Apr 24, 2015, 10:05 AM ]
  • Wondered in a meeting how different views can be used to approach same matter. Marketing, Technology, Helpdesk, End Users, Invoicing, Server Management, Software Developers, System Peripherals, Customer Sectors. And so on. Result? Lot of semi great discussion, but very little factual results.
  • Something different: G6, PzH-2000, M109 Paladin, B-1 Lancer
  • Read: 10 Myths of Enterprise Python
  • Checked out: Apache Flink - It surely looks pretty interesting and something I would use if I would need it. I'll keep this one in mind. - Kw: apacheflink, flink, apache, bigdata, hadoop, inmemory, java, cluster, yarn, hdfs, mapreduce, hbase.
  • Checked out: Google Cloud Dataflow - Hmm, all github examples in Java, checked out but didn't bother to run. Yet I don't see any use for this in near future.
  • Read: Hackers guide to Neural Networks
  • Read: Deep Learning vs Machine Learning vs Pattern Recognition - kw: Convolutional Neural Nets, ConvNets Big-data, PaaS, Artificial Intelligence.
  • Read: Why does Deep Learning work?
  • Plan: I'm going to focus my studies on Data Analytics and Open Data for a while.
  • Had some fun, as usual. With different kind of combination, packaged and recipe articles while doing CMR, BI and ERP integrations. It's one of my favorite topics, because really simple things get so complicated after all. Especially in situations where external stuff gets referenced by parts of the package but that must not be visible to end user and so on. Of course if everything would be done from scratch it wouldn't be so hard. But usually there are pre-existing complex systems and you'll have just to find a simplest possible way to work around restrictions of these systems while still delivering what's required by the client. That's my daily stuff.
  • Took deeper look at Ansible and SaltStack (Salt) to evaluate those again, tried a few things in test environment. Windows Remote Management (WinRM) PowerShell (ps) - kw: configuration management and orchestration tools, playbook, execution module, state module, Python, SLS, YAML, ZeroMQ, SSH, inventory, orchestration, Vagrant, mocking, testing, scripting, modules, events, reactors, JSON, Linux, Windows, Tower, CLI, templates, master, minions, PyDSL, Jinja, desired state configuration (DSC), Powershell.
  • Read: Moving away from Puppet, SaltStack or Ansible?
  • Read bit more about Vagrant - Why they say development environments? Does development and production environments have some kind of difference?
  • Lol, one guy said in his blog "I spent my vacation reading computer literature and documentation". That someone is like me. ;)
  • Configuration Management (CM), requires quite a bit of server fu, duh. There's just so many dumb ways to fail. Is Ansible or Salt even required, PowerShell is the core of Windows configuration, does DSC do everything required?
  • Checked out Azure Regions and Zones as well as Microsoft Azure Regions and Azure RemoteApp. - Again costs are major player with this technology.
  • PaaS doesn't relieve you from software maintenance. Technologies which are used in your apps will be shut down and you'll need to migrate your apps to utilize newer solutions. Google App Engine Master/Slave datastore will be shut down in July 6, 2015. Applications must be migrated to use newer database (NoSQL, HDR or NDB) before that. I think I'll just shutdown my old projects because of this. I don't have interest of migrating those. kw: datastore, Paxos, eventually consistent.
  • Checked out bunch of HTML5 Front-end frameworks. It's horrible, there are just so many options, in hundreds easily.
  • 9ox.net finally closed, because google deprecated and shut down the database it was using.
  • Had a long discussion about different webshop platforms with my friend which is working in the field. - Prestashop, Drupal, Magento, Virtuemart, ZenCart, osCommerce, Wordpress ja OpenCart.
  • Had some discussion about developing electronic purchase receipt formats further.
  • Watched Haaga-Helia Future Forum marketing and system integration videos including TARU videos which is about future digitalized and integrated business systems. WhatsApp marketing. XBRL, Real-Time Economy (RTE).
  • Enjoyed wondering different Microsoft Windows CAL options.
  • A Great slideshare post The Emerging Global Web - how Internet is changing the global trade. There are winners and there are losers.
  • Read: Building Two-factor Authentication
  • JSON API - A standard for building APIs in JSON. - Nothing new at all, but it's nice that there are good examples for people making their first JSON APIs.
  • I was asked if I want to join a SIG developing standards for business message formats for Open Data (sorry, no details at this point) and integration APIs. Well, that's most interesting and yes. I'm naturally interested to do that. I just believe my approach to many problems is really pragmatic. Let's see how that works out with people who got much more theoretical or academic approach. I guess it's good and brings out differences and can lead to valuable outcome.
  • Read: Startup advice briefly by Sam Antman - Too short? Maybe it's better to check out the long version.
  • Radical Statements about the Mobile Web - More web vs native apps discussion. I personally don't want to install any apps or crap on my phone or on my computer, unless I know it's absolutely vital software. I always prefer web over native apps in most of cases. But that's just me.
  • I got involved in discussion if using 2FA will stop hacking or drastically improve security. Well, it will projetect from SOME attack scenarios but it surely won't stop hacking. Sure. So usually site hacking is done via exploiting some vulnerability on site or system. At that point they can usually steal anything they wish from the system. So using 2FA won't actually protect against that. As well as if they truly rooted the system, they can do what ever they want. As well as circumvent password protection measures as well as access all data, source codes, and so on. At that point, I really don't care if you got my password or not, that's least of my worries. My old password to Slack was: Q-CfK4h1H_bB0mN7PPvD I guess you or nobody else doesn't care about that fact at all. More important than slack passwords is the data in the system, like if people have given credentials to other systems during chat conversions and so on. From the end users point of view, if they have rooted your device, 2FA won't help either. They can steal already authenticated cookie, they can route traffic via your device so IP won't seem strange, as well as they can basically do what ever they want to. So, yes 2FA protects from some threats, but it really won't protect you from hacking at all. I could go on about this for much longer, but I believe I made the point clear.
  • Lightly checked out: Apache Kafka, Apache Samza
  • Turning the database inside-out with Apache Samza - Aww, so much talk about things which are obvious to everyone. Replication, caching, indexing, transactions, race conditions, locking issues, materialized views, data transformation, replication stream, transaction log write ahead log, immutable facts, immutable events, better data, analytics, fully precomputed caches. It's better to skip straight to "Let's rethink materialized views!" section. Other keywords: HTMLDOM, JSON, CSS, React, Angular, Ember, Functional Reactive Programming (FRP), Elm, Publish, Subscribe, Notify, Request, Response, Meteor, Firebase, Subscribers, RethinkDB, Designing Data-Intensive Applications, stateful steam processing.
  • Created: Google+ Brython Users Community
  • Studed several posts from Open Knowledge Blog