hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Edward Capriolo <edlinuxg...@gmail.com>
Subject Re: Why windows support is critical
Date Mon, 01 Mar 2010 17:11:18 GMT
On Mon, Mar 1, 2010 at 11:47 AM,  <nocones77-groups@yahoo.com> wrote:
> This is my first post to the group, so I'm not sure I have a lot to add to the conversation
yet. But I've been lurking/searching for a week now, and wanted to add a "me too" to Ravi's
comments.
>
> The quick-start would be fantastic if it actually worked cross-platform and did something
meaningful. But I've been very
> frustrated by the problems just doing development on Windows 7 x64,
> which appear to have been experienced by others too but not fixed. Had I understood how
much of a pain this alone would be, I'd have contemplated re-installing my OS, but this didn't
become apparent until after several days of struggle, and hindsight is 20/20.
>
> I ended up creating a pseudo-distributed installation on Ubuntu in a
> Virtual Box. It all works fine from localhost, and I can run the shell. But I don't see
how that's useful to anyone who actually wants to build a real application. I'm struggling
to
> figure out how to "connect" to it from a remote Java client. The documentation in this
area seems to be sorely lacking, and the painfully brief "Also note:" comment in the quick
start doesn't seem to actually work. (preemptive comment: yes, I know Virtual Box and can
connect to the Ubuntu machine in every other way I've tried)
>
> Unlike Ravi, I'm confident that the time investment will pay off, and I am viewing this
as an opportunity to really understand what's going on under the covers. I've been building
large-scale (mostly SaaS) applications in Java for
> over ten years and I'm very comfortable with Linux. But it has still been difficult to
get started with HBase. I'm making progress...but VERY slowly. Like Ravi, I'm also working
by myself, attempting to start something new, and need to show progress ASAP.
>
> Anyway, the bottom line is that the initial learning curve of
> "bin/start-hbase.sh and start playing around" turned out to not work cross-platform,
and wasn't very very useful, even when it worked. I've resigned myself to having to understand
> almost everything about the entire Hadoop+ZooKeeper+HBase architecture just to do a meaningful
"hello world" from a Java client. I
> understand why some might give up...
>
> Neal
>
>
>
>
> ________________________________
> From: Ravi <ravikapoor101@gmail.com>
> To: hbase-user@hadoop.apache.org
> Cc: Tim Robertson <timrobertson100@gmail.com>
> Sent: Sun, February 28, 2010 5:24:48 PM
> Subject: Re: Why windows support is critical
>
>
> Tim, the only thing I can say to that is that the project needs to be
> able to grow to the size where it will become EC2 capable.
>
> Essentially I represent 1-2 person startups in garages where time to
> market is probably one of the most important criteria. Environment setup
> time and learning curve are factors against widespread adoption for any
> software.
>
>
> On 2/28/2010 11:54 AM, Tim Robertson wrote:
>> For 2) if you look at slides 15,16,17 of
>> http://www.slideshare.net/ghelmling/hbase-at-meetup you can see there
>> are some nice higher level APIs popping up for HBase that help you
>> deal with Objects and not Byte[].  That is open source and can be
>> pulled from http://github.com/ghelmling/meetup.beeno which might be of
>> interest.  There are others floating around as well that might be
>> worth checking out.
>>
>> Isn't Windows used only for .Net development? (;
>> I'd have thought that if a project is of sufficient size to consider
>> HBase, then it is big enough that people would set up appropriate dev
>> environments (Mac, Linux, or windows and use EC2 or something as the
>> dev/test cluster).
>>
>>
>> On Sun, Feb 28, 2010 at 5:17 PM, Ravi<ravikapoor101@gmail.com>  wrote:
>>>
>>> I understand that and I agree it cannot be used on windows in production.
>>> However
>>>
>>> 1. I am talking about windows as a development environment.
>>> 2. I am talking about ease of use of API provided
>>>
>>> These are two critical needs of any web application developer. Don't you
>>> think?
>>>
>>> Ravi
>>>
>>>
>>> On 2/28/2010 11:09 AM, Jeff Zhang wrote:
>>>>
>>>> But HBase is build upon hadoop which is not supposed to run on windows as
>>>> production environment.
>>>>
>>>>
>>>>
>>>> On Sun, Feb 28, 2010 at 8:01 AM, Ravi<ravikapoor101@gmail.com>    wrote:
>>>>
>>>>>
>>>>> Hi everybody,
>>>>>
>>>>> I want to bring up a critical issue i.e. windows support. Today I decided
>>>>> to move away from hbase to mongoDB and I want to explain why and I hope
>>>>> this
>>>>> will help the hbase team understand needs of users like me and ultimately
>>>>> a
>>>>> better product.
>>>>>
>>>>> I am an entrepreneur working on a web application. I do not have big
>>>>> money
>>>>> or big team. So while I am trying to move a mountain, I want the path
of
>>>>> least resistance and want a DB that does not pose additional challenges.
>>>>>
>>>>> Ultimately I found hbase to be more difficult to install and API to be
>>>>> difficult to use on my windows machine (where I am doing the
>>>>> development).
>>>>> After struggling for a few days, I realize that I am losing steam and
I
>>>>> better choose a path of least resistance and I am switching to mongoDB.
>>>>>
>>>>> I do think that hbase is currently ahead. Mongo DB is missing a few
>>>>> critical features for a large website. However if I lose steam, there
>>>>> wouldnt be a website.
>>>>>
>>>>> So I request the hbase team to think if users like me and windows
>>>>> development environment is important enough for the team. A one click
>>>>> install on windows and mongoDB like driver would go a long way to compete
>>>>> with other DBs.
>>>>>
>>>>> I hope that helps
>>>>> Ravi
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>
>>

I can understand the frustration of setting up working functional test
environments. I have found that even with a LAMP application getting a
functional test environment can be difficult. In particular if you
have large fast moving data.

I also understand your struggle. I hobbled with getting my IDE happy
with hadoop/hive/hbase classpath. Also job launching. As you have
probably learned the hadoop "stack" does not really have a "client".
If you have the correct network access, the correct configuration, and
the correct binary you have yourself a "client" and more actually.

As for start-up problems, I can imagine most things have the same
cost. If you do not want to struggle through something, you can find
people in the know that can give you a big leg up.

For example I have a CCNA and know networking well. But if someone
dropped a cisco pix firewall on my desk, it might be a month before I
get that firewall into production. However there are people that could
accomplish that same task in a week, or possibly days, because they
are in the know.

The same is true for hadoop/hbase I believe. If you tool a standard IT
pro/coder and then said ok we need hbase, they might struggle down
some wrong avenues. If you got someone in the know to talk to they can
help you avoid slowdowns.

On this issue of windows, in particular I would have strongly
suggested that the hadoop stack has many linux-isms, and would have
suggested training VM's or some type of linux terminal server for
development.

Mime
View raw message