next »

[isthisthingon]

Hey perk, I was thinking about our conversation today and thought the Ubuntu Enterprise Cloud might be worth looking at: http://www.ubuntu.com/cloud/private.  I have no idea if it's solid or not, but it might be a nice open source mini-cloud to evaluate. 

Here's the preface from that large cloud document I posted earlier.  It's helpful to think about what a private cloud might be capable of providing before really jumping into why to adopt any of these technologies 


Preface

Our society has become dependent on the delivery of large number of complex IT-based services to users
and customers. To support the delivery of services to consumers, the service providers – whether businesses
or government – have traditionally employed various client/server architectures where the client software
can issue service requests to the server.
An outstanding problem of client/server architectures is that demand for services is not constant and,
occasionally, hard to predict. As a result, to assure continuous service availability and quality, contemporary
service infrastructures over-provision resources to support peak demands, which imply high costs to service
providers and consumers.
Various solutions have been proposed to alleviate this problem since the late 1980s. However, in spite
of significant advances in IT, the state of the art in service-oriented computing still calls for efficient implementation.
In particular, the critical ingredient of service-oriented infrastructure, needed to facilitate the
delivery of continuous and reliable services, is yet to be realized. The RESERVOIR project is an aggressive
EC-funded research initiative to address this challenge.
The high-level objective of the RESERVOIR project is to provide a technology foundation for a internetscale
data center where resources and services are transparently and flexibly provisioned and managed like
utilities (zactly - itto). Specifically, RESERVOIR will introduce a next-generation IT infrastructure for deployment of
complex services across different administrative domains, while assuring QoS and security guarantees.
The main ideas underlying the RESERVOIR vision are:

• Provisioning Services as Utilities, envisaging the delivery of IT services delivering IT services, on demand,
  at competitive costs, and without requiring a large capital investment in infrastructure. Our
  research aims to enable the delivery of IT services to the common day public services such as electricity
  and telephone.
• Service and Resource Migration without Barriers, extending service migration capabilities available in
  today’s offerings to work across a large geographically distributed infrastructure that spans administrative
  domains and software/hardware platforms.
• Federated Heterogeneous Infrastructure and Management, introducing a layer which will provide for
  generic management of virtualization and grid technologies, thereby enabling the federation of disparate
  infrastructures and the above-mentioned capabilities of ubiquitous utility-like service delivery
  and migration.
• Advanced End-to-End Support for Service-Oriented Computing, addressing the fundamental technical
  issues of service computing, such as end-to-end security, service deployment and management, service
  billing, and interpretation and monitoring of Service Level Agreement (SLA) conditions.

With those capabilities, the RESERVOIR project aims to support the emergence of the Cloud Computing
paradigm, where services are software components exposed thru a network accessible, platform and
language independent interfaces, which enable the composition of complex distributed applications out of
loosely coupled components.
The prime deliverable of the project is the architecture – presented at length in this document – and a
reference implementation of a service-oriented infrastructure which, building on open standards and new
technologies will provide a scalable, flexible and dependable framework for delivering services as utilities.
We will demonstrate how this infrastructure supports the deployment of several complex service scenarios
that are not otherwise supported by contemporary environments. In doing so, we aim to achieve quantified
and significant improvements in service delivery productivity, service quality, service availability and service
dependability.


 

[perkiset]

ehPHUQUE here comes an Advil binge, I can just feel it.

[nutballs]

So which cloud is this? I get so freakin confused reading the marketing babble that I can't for life of me figure it out anymore.

I just want to add N-machines to the cloud, and have N-apps/databases/websites/whatever running on it, distributing as needed.

The problem that I keep seeing with the cloud is, that either its just:
elastic computing, ie, individual machines, running independently, being reconfigured for the next use when needed.
OR
its a true single machine, with N-nodes making it up, but that makes my head hurt for dealing with a now 300 gig database of mine.

I would LOVE to have a big ass machine, with N-nodes, especially of arbitrary hardware, that when it gets over taxed, I can just plug in a new machine, and yeay, it works, and my machine is now bigger and faster. Is this that?

[perkiset]

This is precisely what I want. A setup where I have one virtual touch point, but can add machines at will to the overall processing/storage capability of the private cloud. ITTO is going to talk about it at the mashup. I just looked at some of the How Tos and nothing jumped out like, "compile this here, setup that text file there and kaboom."

I swear if I get one working I will create the best end-to-end This Is How You Do It that has ever been posted.

[isthisthingon]

http://www.ubuntu.com/cloud/private.  I have no idea if it's solid or not, but it might be a nice open source mini-cloud to evaluate.

Nothing I've seen yet is as plug-n-play as you both are wanting.  In fact, the whole cloud movement is so solidly in its infancy that there hasn't even been agreement on whether an application as a service should be called AaaS.  Ass, that's a fantastic friggin name. 

But all I'm suggesting is to look at the Ubuntu "private" cloud.  It may be close to what you two are asking for but my suspicion is hell no.  Why?  Because each separate element of "as a Service" has it's own unique complexities.  Infrastructure as a Service that's currently available and professionally offered is a public for-pay cloud.  But learning how to do x scaling or y scaling, such as a MapReduce style search across multiple Hadoop boxes is valuable in itself.

But private cloud frameworks are coming and damn that's exciting.