Subscribe: Comments for I'm just a simple DBA on a complex production system
Added By: Feedage Forager Feedage Grade B rated
Language: English
big data  big  comment big  comment  data news  data  error messages  error  messages  news oracle  oozie  oracle openworld  oracle  poc 
Rate this Feed
Rate this feedRate this feedRate this feedRate this feedRate this feed
Rate this feed 1 starRate this feed 2 starRate this feed 3 starRate this feed 4 starRate this feed 5 star

Comments (0)

Feed Details and Statistics Feed Statistics
Preview: Comments for I'm just a simple DBA on a complex production system

Comments for Just a simple Hadoop DBA

Adventures with Data and Massively Parallel Databases

Last Build Date: Mon, 05 May 2014 11:02:28 +0000


Comment on On Error Messages by laurentschneider

Mon, 05 May 2014 11:02:28 +0000

Hello prodlife, what's up? Oracle should implement the F1 key so that we get some really useful help and even online help But I suppose you have no F1 key on your device :) Cheers Laurent

Comment on On Error Messages by rodgersnotes

Wed, 09 Apr 2014 19:59:19 +0000

Absolutely! Good post! I agree, Oracle error messages are usually pretty good; you know the cause and solution pretty quickly. Except for messages concerning networking, which have often lead me one direction, when the solution was actually in another. Car fix it manuals have the matrix: Problem, Probable cause, Solution. ie. Car will not start. Out of gas. Fill gas tank with gas. I've written a lot of thoughts on error messages here: Poor Or No Error Trapping Interestingly enough, so many of the best solutions are not so much in documentation, as they are on individuals' blogs. What if we were to decrease engineers' bonuses by the length of time required for some strangers to figure out their error "messages" ? :)

Comment on Using Oozie in Kerberized Cluster by prodlife

Mon, 10 Feb 2014 23:12:35 +0000

Thanks Karthik, but this blog is not really the right place for "NoClassDefFound" errors. How about asking oozie-user mailing list? Or if you use CDH, perhaps the cloudera community forum?

Comment on Using Oozie in Kerberized Cluster by Karthik

Wed, 15 Jan 2014 17:16:40 +0000

good article... but when i try to do these steps and run the workflow... i get an error in hcatcredentialhelper.. where it is not able to find HiveConf.class.. (NoClassDefFoundError).... i have set oozie.use.system.libpath=true... oozie.libpath=${nameNode}/user/oozie/share/lib what am i missing???

Comment on Three Things To Do Before Starting Hadoop Project by Steve Karam (@OracleAlchemist)

Mon, 21 Oct 2013 15:16:21 +0000

public class OceanInputFormat extends Everything The POC phase is where we are right now and having the toughest time with. The problem is getting business input since they are obviously the stakeholder that will pay for the whole thing. However, when we talk to them about doing a POC the usual response is something akin to "Oh wow, can we get a report that tells us our student sign-ups each week?" Simple stuff we could do with our standard reporting or even a SQL Script. So prior to doing a POC, we're having to change people's perception on what a 'business problem' can be. I think people get so used to hearing "we can't do that" that they forget all the awesome ideas and questions they had when they first came on board. So despite the prevalence of "Big Data for Business" type books on everyone's desk, right now we're focused on doing small projects with IT data...which unfortunately no one cares enough about to pay for.

Comment on Three Things To Do Before Starting Hadoop Project by Ofir Manor

Sun, 20 Oct 2013 18:53:57 +0000

Great stuff Chen! I agree with all points - automation and incremental value are very important. Regarding POC - in a way, you describe building a prototype together instead of a POC. That is almost always better, but the poor customer has to pick a vendor first... A bit of chicken and egg problem... I think that aggressive competitive POC might be great to well-understood domains, like picking best NAS storage for enterprise environments.. For messy, cutting-edge platforms - pick one of the leaders that you trust (and that has proven track of delivery in your region and maybe your vertical) and work together.

Comment on Big Data News from Oracle OpenWorld 2013 by laurentschneider

Mon, 14 Oct 2013 19:22:59 +0000

While multitenant option is pointless (you pay more to save money), there is a real benefit in this one, so I will go for "probably usable in 2016" and "as much as partitioning"

Comment on Big Data News from Oracle OpenWorld 2013 by prodlife

Mon, 14 Oct 2013 19:08:49 +0000

Excellent question. You probably know as much as I do about those :)

Comment on Big Data News from Oracle OpenWorld 2013 by laurentschneider

Mon, 14 Oct 2013 18:44:27 +0000

> let me know what you think of In-Memory I'll love it <3 How much does it cost? When is it available?

Comment on Big Data News from Oracle OpenWorld 2013 by Ofir Manor

Tue, 08 Oct 2013 20:39:07 +0000

I'm sure Oracle will continue to push Exadata... but consider this: - The main DB message of OW is "buy in-memory option and have 100x faster analytics with 2x faster OLTP". - If Oracle will keep pumping the message for a year, and repeat it next OW when the option is GA, some people will take it seriously. - If you have a DB with mixed workload, why buy both Exadata and In-Memory option? 100x + 2x is more than enough for your EBS / Siebel / what have you. Exadata was supposed to solve the analytics bottleneck, and In-Memory is even faster than I/O (both random access and sequential scans...) - As a bonus, with In-Memory Option, you may even run your production apps DB in a big VM in your VMware cluster - just give it 500GB and 32 vCPU, no need for propriety cluster.. Some may be tempted to try it. OK, last point seems a bit adventurous , but why not...