BigDataWeek

you don't need to look ahead 3 years for that EMC's Isilon is doing it today!

I agree storage should be "smart" and take data from the apps and be more "horizontally" integrated

@DBAStorage maybe send a calendar invitation for 2/19 announcement?

.@vStewed Shared storage takes away adv of Hadoop. But a hybrid model may be needed.

@DrZeydy true scale-out storage platforms can minimize the architecture impact

Dynamically deploying nodes to jobs is easy - managing (storing, granting access, protecting, etc) TBs or PBs of raw data is hard (aka #DataGravity) thus my perspective

compute is the easy part. Persistence still remains challenging the Data Center.

I think it depends on how real ServerSAN is - which I realize you don't subscribe to

EMC will make a public announcement on February 19th, 2015 highlighting our ongoing commitment to keep Big Data secure and accessible for archiving and analytic needs. https://community.em...

I agree value in ServerSAN but disagree with inclusion of hyper-converged in ServerSAN category

@DBAStorage making the #DataLake viable is key. #RedefineDataLakes

certainly agree that securing the hadoop ecosystem is a key requirement for its enterprise success

you lose flexibility with HCIA as it is today. Maybe a new approach will be needed.

as a new platform emerges, a more flexible approach to HCI will be required for #BigData

I don't get it. The biggest disruptive factor for emergence of Hadoop is cost of storage. How can you get cost advantage with shared storage.

with the proper storage architecture. Not your Enterprise High Platform.

HDFS on DAS was a great way to break barriers and start Hadoop pilots, though

Much of the ROI on Hadoop has been "Reduction on Investment" - quote by my good friend @ab_hi_

data reducing shared storage can require a fraction of the raw capacity over multi-replica DAS while providing native access to other platforms

also on protection efficiency. 3 copies are expensive at scale.

Apps have to be born adaptive -- adjust functionality to a given data context.

seems like apps will be developed for Hadoop or alternative access Hadoop data from a shared storage backend. The latter seems to have less lock-in.

good point. Some reading here: http://research.giga...

Interesting stat - the 2 most critical tools in #BigData projects: 1/ the existing EDW and 2/ Data integration tools

Collecting a bunch of unstructured data in one place is borderline useless unless you have applications that can take advantage of the information that data warehouse inherently provides.

never confuse data with information

Exactly what I was calling attention to. :)

The #DataLake works for those who can leverage it.

@vGazza no it doesn't #DataLake is an EMC marketing term

@vStewed it is not. I just read Price Waterhousecooper leveraging the term.

@vStewed take a look here: http://www.pwc.com/u...

@vGazza Thanks, I understand how marketing works

@vStewed it is no magic, it is hard work. #RedefineDataLakes

- appdev shouldn't care, the ultimate goal is to keep the infrastructure out of the way of progress.

I look at #cloudera what #netscape was to dot com boom

what do you mean - do you see them crashing or being the enabler to a bubble or boom?

interesting point of view and great question!

if Oracle and Cloudera don't join the Open Data Platform, they will become one! :)

enabled. Netscape crashed due to other factors but the change they brought to market place was permanent. In the same way the big data movement which Cloudera started is going to be permanent. I hope they reap most rewards from it.

Kudos to @Cloudera for disclosing - @Veeam did the same a few weeks back. A new trend?

controlling the narrative and counterstrike to Pivotal Hortonworks news- smart strategy

@DBAStorage maybe that's the plan #justsayin

thoughts on Cloudera Revenue announcement - what % might be Software vs Services/Support?

Cloudera blew thru $100M in revenue - not a huge surprise but not much more in that announcement - Like Dell - cloudera can cherry pick financial data (for now)

@vStewed u guys next? What data did they share though really - very common for pre-ipo no?

Netscape (Jim Clark) totally underestimated the impact of MSFT bundling the browser into its OS

@dvellante Controlling the narrative is important - but what's the counter-narrative being pre-empted here?

#apachespark

to be honest the only other compute technology we see being adopted by clients is #clouderaimpala

Spark is very relevant and great for real time analytics; the innovation beyond in memory - is in-processor that's the real deal #silicon

when memory becomes non volatile at scale, I/O will not be limiting anymore

There's so much runway here it's silly. Let's get the world from disk to flash before we talk "in-processor"

Ever since the dawn of memory there's been talk of "in-memory" - it's more real now than every because of costs, etc - still have to protect the data

@dvellante the greatest potential for innovation is in the frontier of memory and storage.

compute will keep getting closer to the chip but at present when using memory is most practical #apachespark has taken persistent storage to memory which itself has brought tremendous information gains

@vStewed the joke the other day was spinning disk will die b4 tape

I agree. I think spinning disk will be dead for enterprise storage. It will stay alive for consumer desktops though

@furrier I agree the disk replaces tape chatter was illogical due to power requirements... tape turns off!

not even desktop storage ... flash wins

@vStewed tapes travel faster than disk too. On cargo.

Not anytime soon. $/GB + online access will keep spinning disk around for a while

@vmtyler for a while. But growing flash density will make a big influence here.

and spinning disk density is growing too

@vmtyler agree. Flash density is behind but growing faster. It will be fun to watch. #MediaWars