Auto-Parallel Technologies, Inc.Services :: Cloud Setup & Installation :: Overview
Data Cloud. Compute Cloud. Any Cloud.
Auto-Parallel Technologies offers setup and installation services for machines running on the Cloud.
Are you ready to upgrade your IT infrastructure by adding Cloud computing resources?
We can help!
But first, we must ask a critical question... What is the Cloud?
In today's technology-driven economy, everyone wants to be "on the Cloud", but very few actually know what
"the Cloud" truly represents. Briefly, the concept of "the Cloud" has it's roots in the ill-fated idea of
"the Grid", which was popular only among high-performance computing enthusiasts between 2001 and 2008.
From 1990 to 2000, the two main ways of building a high-performance computer were all-in-one supercomputers
(prohibitively expensive) and "Beowulf" clusters of normal Linux computers (cheaper but slower), both of which
employ multiple individual processors operating in parallel.
A few large organizations, mostly universities and computer companies, owned multiple high-performance computers,
along with large numbers of normal desktop computers; one such university could then combine all their collective
resources to form a "Grid" computing system, which essentially operated as a single, huge, heterogeneous
high-performance computer. Significant Grid software control systems included Globus, Legion AKA Avaki, and Sun
Grid Engine.
Two distinct categories of Grid system were defined: "data Grid" and "compute Grid". A data Grid simply served
as a decentralized network-attached storage (NAS) system, allowing users to store and retrieve large data files.
A compute Grid performed the much more complicated task of serving as a decentralized high-performance
computing cluster. Proponents of these technologies envisioned "the Grid" as the world-wide Grid-of-all-Grids,
the cumulative computing power of all individual Grids on Earth combined.
The original promise of the Grid was
"high-performance computing applications served like electricity"
, which could also be re-worded as
"supercomputing power at your fingertips"
.
Although this was a grand and welcome idea, the near-total lack of publicly-available parallel software applications
essentially ended the life of the "compute Grid" before it even began. Thus, the majority of "Grid users" were
only data Grid users, which was a trivial use of expensive Grid systems when compared to the failed promises of
compute Grid technology.
The ultimate demise of Grid technology came with the increasing complexity of managing an in-house Grid system,
when compared to the simplicity of out-sourcing the task to a 3rd-party network provider hosting company, such as
Amazon. Post-bubble technology infrastructure investment and the economies of scale gradually lead to remote
Grid services being offered with utility-like pricing, halfway fulfilling the original Grid promise. Due to the
commercial failure of Grid technologies, savvy IT marketing executives quietly re-branded "the Grid" as "the Cloud".
(Not long before, the same IT marketing executives had re-branded in-browser AJAX web applications as "Web 2.0", a
buzz-word which quickly became over-used and was subsequently discarded by the Internet community.)
Thus, we arrive at a clear (if somewhat difficult-to-reach) definition:
"A Cloud is a Grid offered at utility pricing rates."
... or more specifically ...
"A Cloud is a redundant remote network of virtual computing resources, likely decentralized and heterogeneous,
offered at utility pricing rates."
The old, simple-purpose "data Grids" are now "data Clouds" containing "big data", and the powerful but
neglected "compute Grids" are now "compute Clouds". "The Grid" is now "the Cloud", the nebulous world-wide
Cloud-of-all-Clouds.
Unfortunately, we are still suffering from the same lack of parallel applications which doomed compute Grid technologies
over a decade ago. Because of this, most Cloud servers are still limited to the purposes of storing large
amounts of data or, even more trivially, simply serving as web servers to host corporate websites.
It is the primary purpose (and namesake) of Auto-Parallel Technologies to solve the issue of parallel applications
once-and-for-all, with the release of the auto-parallelizing RPerl compiler software.
Soon, more and more compute Cloud systems will be powered by RPerl,
finally fulfilling the full Grid promise.
To demonstrate the power of an RPerl-driven compute Cloud, APTech has developed and launched the new
CloudForFree.org platform.
All Cloud platforms must, by definition, provide the following:
- Virtual Machines
- Remote Accessibility
- Physical Hardware Redundancy
- Data Storage and/or Computational Cycles
- Utility Pricing
