Google IO 2013 #io13 – Voice dictation and Chrome Android convergence.

by: Philippe Back - May 16th, 2013   Filed under: Improvement, Tech Add Comment

Google IO logo

Yesterday I was attending the Google IO event. Pretty interesting. Now I’ve been using some of the technologies that were presented. I am a pretty heavy user of Google applications and find them quite useful but they’re still in need of more integration.

I’m pretty sure convergence between Android on Google Chrome is the key for an advanced user experience.

There’s nothing as a single form factor that will please or be suitable for all needs and use cases.. I’m now sitting in front of a 24 inch monitor, there is also a 10 inch tablet and this 4.5 inch mobile phone (Galaxy s2 plus, why mention Samsung anymore, they are Android phones). Obviously you can do much more with some real screen estate than with any phone but the phone proved to be quite a reactive to capture events whereas the bigger boxes are useful for creating interesting deeper content. Mobes are useful to capture moments and record stuff. For exchanges and for crafting content in communityies these things are really useful. Still, my concern is that you need to learn and try these things. It is not sure that this provides you with a productivity boost because all the time your gain from using the tools if offset by the time you need to learn them. Just because of the nerve wrecking rate of change that is pushed pushed pushed on you. One cannot use all of the these things.

I’m trying to type this by using the Chrome Speech API.

Something like that is pretty neat but its not integrated so I have to go to a ctrlQ.org for a dictation application and copy paste in Google Drive, switch the language in Chrome to get it working well.

I actually do know what to do to get it to work. But is anyone going through all of these moves to get things working? That’s a pretty wide chasm to cross if you ask me.

So, I am now in the process of digesting all those new ideas and products. I wonder how they do monetize all of this as there wasn’t that much about ad placement. It is strange to me as Google is making money that way. What I saw was more competition kill moves. I am curious to know more about their strategy behind the scenes. One thing is clear, competition heats up for Google too. 2013 will for sure be very interesting to watch.

Speaking of which, you can follow sessions at the live stream

Logging MySQL queries with MySQLProxy

by: Philippe Back - February 22nd, 2013   Filed under: Improvement, Tech Add Comment

In need of viewing what went on the wire

I have been struggling with figuring out why something was wrong in my SQL queries.

I needed to see what was happening in real time and also keep a little log of things.

My MySQL server has the logging enabled, slow queries and all. It even logs to TABLE.

But this wasn’t doing the trick.

MySQLProxy

Enters MySQLProxy.

MySQLProxy mediates between your client app and the MySQL server (or servers, consolidating a bunch of servers into one, along with the ability to set a bunch of them as read sources, and another as a write source — which for scaling appears pretty smart).

More on the scalability ability for another time.

What I wanted is to log the traffic.

Getting MySQLProxy

Easily enough, get it.

I installed 0.8.3-Win32 as my dev box is a Windows 7 system (64bit).

Configuring MySQLProxy

First of all, MySQLProxy will listen by default to port 4040 and forward to 3306, the ususal MySQL port.

What needs to be done is to provide the host details as a place for the thing to log its technical output (to figure out what goes wrong).

I wrote a script for handling the startup: runmysqlproxy.cmd

set MYSQLPROXY_HOME=%programfiles(x86)%\MySQL\mysql-proxy-0.8.3-win32-x86
set MYSQLPROXY=%MYSQLPROXY_HOME%\bin\mysql-proxy.exe

"%MYSQLPROXY%" -b 127.0.0.1:3306 --log-file=E:\var\logproxy.log 
    --log-level=debug --proxy-lua-script=D:\Dev\tools\sqlproxy\logquery.lua

Logging queries

It happens that MySQLProxy can do pretty much everything you can dream with queries that do pass through. Okay, I just wanted to log queries to a file.

How to? Lua scripting to the rescue! Lua.org

That’s why logquery.lua is for.

Let’s check it out:

function read_query(packet)
   if string.byte(packet) == proxy.COM_QUERY then
     query = string.sub(packet, 2)
     log_query(query)
   end
 end

 function log_query(query)
 output = os.date() .. ";QUERY;" .. query
     print(output)
 local file = io.open("E:\\var\\mysqlproxy\\queries.log", "a")
 file:write(output .. "\n")
 file:flush()
 file:close()
end

I really never coded anything with Lua but it for sure will happen due to the power of that proxy.

Connecting to the proxy

Just a matter of using the mysql command line client (with my very secure credentials…):

mysql -hlocalhost -P4040 -uroot -proot

And then issue some queries.

These are captured:

02/21/13 23:45:11;QUERY;SELECT DATABASE()
02/21/13 23:45:23;QUERY;select * from tiki_todo

And the file filled in.

Hope it will help you get started with MySQLProxy.

Going further

There is quite a bunch of possibilities as you can see from the scripting reference.

Happy MySQLProxying!

Microsoft Innovation Center Boostcamp: 15/10 – Mark the date

by: Philippe Back - October 9th, 2012   Filed under: Business, Tech Add Comment

Next week, the MIC Boostcamp has another session.

I have volunteered as one of the “experts” (well, listening to my advice usually helps in improving your strategic position, makes more money come your way, and reduces your labor intensity — what’s not to like in that?).

See you there. As I’ll be there all day, feel free to send me a tweet or two if you want to discuss matters (http://twitter.com/philippeback)

Websphere or Open Source (Free is not Free) at #IBM #ICTY by @sebsto

by: Philippe Back - October 2nd, 2012   Filed under: Tech Add Comment

Interesting topic as far as I am concerned.

“Libre et Gratuit”

Sebastien Stormacq is our speaker. Pretty decent english anyway.

Here is our plate: IBM and FOSS, AppServer Landscape, TCO, Differentiatiors, WebSphere family, Fast and Lightweight.

Well, quite a heap to go through…

IBM loves open source

Of course. Linux Kernel, Apache, Eclipse, Open Office (and Lotus Symphony will be open sourced).

1000 devs on IBM payroll contribution to FOSS projects.
150 projects with IBM in there
80 of which are led by IBM

IBM is giving, but IBM is consuming as well (fair game, right)

Websphere integrates quite a bunch of FOSS stuff.

Note to self: Wind, Tuscany: need to check those

The SOA reference architecture in OSS

The SOA ABBs.

Open Source has solutions for all of the blocks. But made by different people and different styles. You’ll have to do the integration and gluing yourself.

IBM does the integration in its labs and you get properly glued shit. So that you can focus on the real business and not tubes and skunkworks crap.

A tour of the App Server (AS) Industry

The usual suspects: JBoss AS, IBM WAS Express, Tomcat, Oracle Glassfish, Weblogic, VMWare new stuff.

Of course, it all depend son what you want to deploy. Java, Groovy, PHP and other things can run on a JVM.

All kinds of workloads as well.

Main small and medium size projects go Tomcat 7.
Sometimes, we go JBoss.

Good alternative: Websphere Base Edition (WAS Base)

And also WAS ND (Network Deployment) for larger thingies.

Apache Tomcat

Yay, the great little server that I do know inside out, code base and all. That’s what you get from training a ton of people for months on end on how to use it.

On the monitoring side, they do tell us that it is pretty much low. But in fact, no. It is full of JMX support. Ah, liars…

There is no paid support there. Hey, this is a huge opportunity for me :-) I’ll create a business unit.

There is now TomEE, which puts a bit of EE stuff. But, forget that, pack in some Atomikos from Dr Guy Pardon and you’ll have a perfectly good JTA implementation that will kick ass and space CPU cycles.

VMWare has tc and spring bundled on top of their vFabric (new ESX moniker). I’d have to check this with Dale Hendrichs as he may be involved. Yet again, a Smalltalker :-)

JBoss is developer centric strategy. It is a RedHat business unit. Good thing these is drop and XML and get resources configured. It now has JEE compliance. The BRE is in there. An advantage. But who really needs a rule engine given the poor state of the general coding community.

JBoss dev is not driven by RH.

Free is not free

Yeah, free is free if your time is worthless. Consider that the acquisition cost is only a tiny part of the whole picture.

Ha! The famous iceberg with 90% hidden. TCO is a deep beast indeed.

All right, IBM shall provided answers to the 90%. There even is a calculator. I am curious to do that.

Obviously, IBM is lower on the TCO. And why not? There is a huge investment they do, thus it is logical that it works better on their platform.

Free is free as long as your time is worthless. Time being the great equalizer, get me people who can implement all of that fast so that we can focus on delivering value. Watering the plants of the whole building for a year is cheaper than the starting cost here.

Development licenses for Free (yay!)

Devtools for Eclipse: free
SmartClouyd for dev: Free (must try)

Cost reduction strategies

From 30% cut down to 80%

Well, looks like competitive on the pricing front.

IBM Workloud deployer, a private cloud solution

Quickly mentioned

Main differentiators for WebSphere vs others

  • Stability
  • Management
  • Monitoring
  • Operations support

Websphere AS Resiliency

WebSphere WAS 8.5 ND (Network Deployment) – New stuff, same price. No need for a ton of packs.

Application Edition Management highlights

  • Upgrade without interruption
  • Multiple editions in parallel
  • Rollout policies
  • Admin console and scripting support

Well, plenty of new stuff to throw at our sysadmin for sure :-)

Health

Superior visualizations

Dynamic Clustering

On demand routing

  • Priority to business critical apps
  • Routing tiers know about server health
  • Route to best server (nice!)
  • Preference for high priority requests
  • ..

Cross component trace

  • Trace log entries across multiple threads/Componets
  • Request ID attached to every request and propagated (that’s we need of non repudiation!!!!!! Worth my day alone!

Performance

SPECjEnterprise 2010

WAS 8.5 beats the crap out of other WAS versions.Not sure about the other products from competition… But they’d like to compare. The competition doesn’t publish stuff.

Summary

Advanced management and monitoring
Health mgt
Dynamic Clutering
Performance

and

Perfect (well, almost :-) ) integration in the WebSphere family.

There is even a migration tool

Check that out to pick up your apps and deployment descriptors.

Webservices look like faster (3x) on WAS.

Yet another difference

Extreme Scale in memory cache. Terracotta from IBM then :-) Or Memcached. Well, anybody needs slabs. That one or another, you need one.

WAS HTTP Session caching benefits from this.

WAS has always been pretty cool on session management and that even since version 3.0 (yeah, I suffered through that old crappy beast using an X-Windows based GUI…)

Very low latency. Much appreciated by banks, right. HFT? HFT and Java? Duh, not really :-) Or compile the Java into Native. That you can do. Get this: (http://www.excelsior-usa.com/articles/java-to-exe.html) And then jump!

Cloud deployment

There is hell of a lot of support for this as well.
You can use a “Cloud Management Appliance” that will help in deploying. So, autoscaling and failover.

For private cloud, works on VMWare ESX (which we use), z/VM (which I do not know about), and Power VM (but I guess you need a Power chip for that. Power 7, mmh, yummy).

Some mobile touch

  • Full development and management environment in the form of Mobile Foundation (see a previous post on mobile)

WAS Liberty Profile

Lightweight, Simple, Fast. Is that WebSphere? Huh?

For developers, this profile attempts to buy back some love.

Released in June 2012.

It is a subset of the full WAS. And it puts the developer back in the driver seat (it was about time).

Profile:

  • Subset of JEE6
  • Web profile (JSR316)

So, IBM has defined the “Liberty Profile”

Let’s test drive

40 MB download
5 secs to launch with zero app
1 configuration file

No admin console, just a file (looks like server.xml)

Dynamic runtime

JSF or JSP requirement detected, so will be launched accordingly.

Platforms for development

  • Windows
  • Linux
  • OSX (in the works, booh, I want one)

And an Eclipse plugin to manage the server.

Fidelity to Full WAS

So, if it works on Liberty Profile, it will run on the production WAS. I’ll take you to the test here!

And you can use Liberty profile in production

Cool! But WAS ND is nice too :-)
For small apps, it is a decent enough runtime

Take aways

WAS vs FOSS

  • Lower TCO and ready to prove it with the “BVA”
  • PArt of a larger family and ecosystem

All right, I drank the cool aid. Now, back to our regular programming and the drink session. Thanks for the prez, it was interesting, even if you talk to devs, get us some demo running.

Intro to Datapower at #IBM #ICTY

by: Philippe Back - October 2nd, 2012   Filed under: Business, Improvement, Tech Add Comment

Datapower is the IBM Websphere Connectivity and Integration Appliance.

Basically, this is full of custom ASICs and stuff and is fast, easy to configure and not prone to all crap that happens to standard boxes.

Datapower was a company in its own right and was acquired by IBM.

How special is Datapower

On the feature set, it is special:

  • Simple architecture: firmware and purpose built hardware. (firmware is often updated to keep up with novelties)
  • Delivered from the factory with everything you ned to connect to the network and start working
  • All computationally-significant components are sealed with temper proof casing (chips, memory, board and card, signed flash file system). Superfast parsing and crypto engines. Truly in a league of its own. Does TripleDES, AES, RSA1.5 etc.

How does Datapower fares on the market?

Pretty well actually.

“…one of the best selling Websphere products since the acquisition of Datapower inc in 2005.”

Well, what’s in a name? From 2002 to 2012.

XA35-XS50-Xi50-WTX-XB60-Blade-XI52/XB62/XE82-XG45

The firmware is developer since 10 years already. Hopefully, this is now pretty good.

JSON is in there now. Phew, enough of XML :-)

Usage areas and users

Various patterns of usage where sensitive information is transported over the internet.

  • Government (Defense, Agencies and ministries)
  • Banking (Encrypting card numbers)
  • Insurance
  • Retail, Utilities, Power, Oil and Gas

Close to 1000 worldwide installations and growing.

What were the classical business cases?

Classic SOA business case, with complex SOAP structures.
And this means a huge computing effort… crypto, signing, auth, filtering, SLA checking, alerts… All in a single box. How sweet.

So, it saves money and headaches.

But messenging has evolved a lot (with mobile coming on the scene).

New use cases

  • Provide secure data communication across channels and protect them (who uses our services, to what extent, limit this leeches, limit traffic to guarantee QoS)
  • Protection from SQL injection, XSS
  • Consolidate (Transformation, Protocol mediation, Routing, Caching)
  • Manage (simple configuration, policy driven setup, centralized governance with WSRR links)

It also is very suited to the Internet of Things era. It can swallow the storm. Without blinking an eye. Well, a led or two…

Support for mobile devices call for special care on security. Datapower to the rescue here.

The Datapower is then living in the DMZ.

In the DMZ here are the rolls to take:

  1. Secure gateway (web services, web apps)
  2. B2B gateway (EDI aware)
  3. Edge optimization
  4. Side cache

In the trusted domain, here are some more:

  1. Internal security point
  2. Enterprise Service Bus
  3. Runtime SOA governance
  4. Web Service Management
  5. Legacy integration

Each of these roles are best served with a special flavor of the Datapower brand.

Let’s say that the next feature will be to brew coffee and sing God Save the Queen. Quite a few things in there indeed.

And ideal beast for deployment into less than fully trusted network (savage gardens, I’d say)

Useful due to:

  1. Common criteria compliance to EAL4 and FIPS 140-2 Level 3
  2. Tamper proof box
  3. Signed and encrypted file system

As a secure gateway

Some physical characteristics

1 U form factor, 4x1Gbps Ethernet port, 2x10Gbps ports.

Feature set focus

Proxying and enforcement

Protocol support: HTTP(S), WMQ, Websphere JMS, FTP(S)
Format support: XML, SOAP, JSON, PKCS7 (option)
Transformation engines: XSLT, DataGlue – WTX/FFD (option) (use Validation maps and Transformation maps)

Using WTX Design Studio (Eclipse based)

Does things like:

  • JSON outside, SOAP inside.
  • Disallow some countries to see some content before they are allowed to

As a B2B gateway

Some physical characteristics

2U form factor
8x1Gbps
2x10Gbps

More memory, more storage

Feature set focus

  • Partner management functions
  • Enhanced QoS
  • Additonal protocols
  • Addition formats
  • Additional transformation engines
  • And including all of the Security Gateway base

Popular uses in the internal side of business as well…

Simplified lower-touch maintenance (1-2 upgrades of firmware per year)

Fast, less expensive deployment (config only, delivery to production in less than a month)

As an ESB and Legacy enablement point

  • A drop-in ESB. Less headaches!
  • Easily service-enable legacy apps
  • …

As an elastic caching within infrastructure

Meaning? Meaning it offloads processing from the backend side.

How to do this?

  • Optimize response time (cache response data)
  • Reduce server load (cache response data)

There are other form factors as well

In the blade form, in need of fitting that in chassis (XI50b “blade”)

And for Systemz. XI50z

What’s new in 2012?

Datapower Firmware V5.0 highlights:

  • OAuth support
  • OAuth scenarios (3-legged, 2-legged)
  • SLA Enforcement and SLDs (Service Level Defintions, synchronized with WSRR)

New WS-Proxy feature with WS-Mediation Policy. Configuration only. Great! Works! (We have integrated this, with V5 and WSRR 8, alerts also go to ITCAM for SOA in Tivoli with the Datapower agent).

  • Application optimization options, allowing to self balance across a cluster of appliances.
  • Application aware intelligent load distribution

The Websphere Applicance Management Center (WAMC)

  • multi box management in a single place.

Check the resources

  • (http://www-01.ibm.com/spftwarE/integration/datapower)
  • youtube has some

Redbooks:

  • Appliance architectural paterns
  • The programmatic management interface

Good presentation on the features. A bid sad to have had to rush at the end on the new feature set.

The future of #SOA at #ibm #icty 2012 with Gerry Keilly

by: Philippe Back - October 2nd, 2012   Filed under: Business, Improvement, Tech Add Comment

The content in here is my own set of notes and not Gerry’s words

Back to the four trends. Mobile. Cloud. Big data. Social.

  • SOA is built upon clean design principles.
  • All workloads are not created equally.
  • Historical workloads are pretty much predictable.

Social and mobile are different beasts!

Snowstorms! Reactions! Checks! Fast! Often!

Kill the API. Need some cache and policy in here…

Volume and speed of interactions are impredcitable events.

Peak management and buffering are the new skills. Yeah! Sweet. Bare metal and instrumentation are coming back!

Expose your enterprise. Expose your flaws. And the crowd is not going to let you breathe for long.

As we do have hybrid deploypment, cloud and on premises, new skills are needed too.

Connectivity and integration technologies

Universal connectivity.

We are back to not so reliable connections.

In fact, back to the same kind of problems of the good old days, only larger and more acute. scale matters.

Websphere MQ 7.5

Integration with MFT. Build an integrated capability for a rich capability layer that will stand the requirements for the upcoming wave! Ture indeed. I am now working with that tech. It rules, period.

WMQ telemetry. Millions of connections. Price lowered.

WMQ AMS.

MQTT brings MQ into the mobile space. TT for telemetry transport.

So, moving to open source for MQTT for ensring adoption. Good idea Gerry.

Gain reliable delivery over fragile connections. 3G anyone?

Leverage information provided by mobile devices.

Ease of integration with enterprise applications.

Build a smarter planet with messaging optimized for smart sensors and telemetry devices.

… With MQ Telemetry.

Monitor a peacemaker over the network! Cool use case.

Put a lightweight MQTT client on any device. Smartmeters, gas pumps, fridges… Exciting!

Cloud connectivity deployments spanning on and off premise

  • connecting to cloud
  • connectivity between clouds
  • Location independance. Switch where the connection is goig under the covers.

a word on APIs

Businesses opening up APIs is a game changer if you do it right. And kill your business if you get it wrong. Damn right! Apigee is cool btw, check it out!

Connect to a broad application developer community.

Extend your core services with an ability to build capability internally.

BTW Salesforce has a huge event these days. And a huge huge huge attendance!

not having an API today is like not having a website in the 90′s

I love that line!

Your API is an extension of your brand.

You can ruin your brand by having this wrong.

Your API extends the value of your brand.

Websphere CastIron Live.

Enabling rapid and trusted entry into the web economy.

Expose, manage, control, run analytics on your services.

Create > Socialize > Manage

90 day trial. Gotta give it a shot!

Too bad it doesnt lists Belgium in the country list…

Websphere Message Broker v8

Powerful beast. I testify from real world exposure. This thing pumps out an awesome amount of messages per second with an incredibly small number of CPUs.

Extreme scale in WMB: cross broker caching. Agregation. Well, a killer. Period.

Every project and everyprocess needs a proven platform

Yeah sure. But startups go node.js on V8.

A new breed of systems

Systems with integrated expertiseand built for cloud.

Okay, that is the integrated capability. Great but the string attached is vendor lock in. No matter how good the vendor. That is where you’ll need people understanding all of this in you company. Who is pointing at me here? Just joking. Half joking that is…

Well, IBM has the ability to deliver for sure. Are clients going to be up to the reauired maturity level to exploit the tech properly?

Will the culture be able to swallow that huge pill? And nit feel obsolete in the process? People are what makes or breaks things indeed!

Extend > Transact > Optimize

Things do change. But fundamentals do not.

  1. Right design practices
  2. Well managed, well governed services

Embarking with IBM may be the smarter choice. provided you have the money of course.