Dataline, Lockheed Martin, SAIC, Unisys on Tactical Cloud Computing

3 Comments

1. Trey on September 28, 2009 at 10:36 pm

   I do not believe that the Army is ready to rely on a cloud computing architecture at the tactical level. I am concerned that Army leadership may not have a clear picture of the challenges facing the field that must be overcome before we can fundamentally change our computing construct.
   First, the bandwidth of a tactical network cannot support SaaS, PaaS or other cloud computing constructs. If every application is online, how much of that bandwidth is consumed by overhead? Tactical units get less than 80Kbps per device. This is barely enough to support a radio-quality VOIP call.
   Second, even if the bandwidth was sufficient, there are severe latency problems with SATCOM that may also be prohibitive. While industry has made amazing strides in data compression and waveform utilization, one fact cannot be changed. Unfortunately, we cannot alter the speed of light. Imagine trying to edit a document on a server three satellite hops away. Every click and keystroke would take three to four seconds to appear.
   Third, the availability and redundancy of the network is not sufficient to support combat operations with cloud-based resources. In monitoring the tactical network south of Baghdad, I can count on one hand the number of times when there were no network outages in a given twelve hour period. Regardless of the cause, the key concern in relation to cloud computing is: How can an end user be productive when they are disconnected from their cloud-based resources? In some cases, being disconnected even for a minute during critical operations can mean life or death to Troops in contact or those requesting medical evacuation.
   There are a few ways to address these issues in order to make the promises of cloud computing work over a disadvantaged and ever-changing network. One construct would be to decrease our reliability on SATCOM. Wireless backbones provided by commercial line-of-sight radios (802.16, OFDM, microwave and variants) provide a more stable terrestrial network that eliminates latency and increases availability. If a unit could establish a redundant line-of sight mesh with a node connected to a high-speed fiber connection, it could receive the bandwidth necessary for cloud computing.
   Another method is to get the cloud resources closer to the end user. In this model, wherever there is a SATCOM node, there exists server capacity to provide services over a wired or wireless LAN. Frequently used applications and products could be dynamically cached, so they can be accessed even when the WAN is unavailable. The servers keep these up to date by synchronization with hosting servers and peers. Meanwhile, the user pulls resources across the LAN instead of the SATCOM link. Unfortunately, the Army seems to be going the other way with its ‘Global Network Enterprise Construct’. In this vision of the future network, IT services would be consolidated farther away from the tactical user in order to best standardize and control services.
   I am a firm believer in promises of cloud computing. I have built virtual servers in tactical units and have seen the cost-benefit first hand. I have provided entirely web based services for everything from intelligence fusion to real-time collaboration. Unfortunately, I also observed that these applications became less accessible to units with poor connectivity. It is my sincere hope that the leadership in the Army will understand the challenges faced by lower echelon units as we move forward towards the adoption of cloud computing. If the Army can leverage the appropriate technologies at the right levels, then cloud computing has the potential to provide consistent, low-cost, interoperable, and sustainable computing services to the Warfighter of the future.
   Major Trey Blacklock
   Information Systems Management Officer
   United States Army

   The views expressed in this BLOG are those of the author and do not reflect the official policy or position of the Department Of the Army, Department of Defense or the U.S. Government

   Reply
2. Kevin L. Jackson on September 28, 2009 at 10:56 pm

   Major Blacklock,

   Thank you for your excellent comments. Your concerns are more than valid for reach back, or "enterprise" cloud computing construct. That is more in line with what DISA is providing with Rapid Access Computing Environment (RACE). "Tactical" cloud computing, on the other hand, uses cloud computing techniques and approaches in order to increase the efficiency and effectiveness of more "local" compute and storage resources (think in-theater). As you allude to, this is more along the lines of linking together the compute and storage resources of five Humvees within UHF range of each other. Another concept would give the fleet commander the ability to link a battlegroup's compute and storage resources together into a single supercomputer in order to run multiple strike package simulations. I think we're in agreement.

   Reply
3. Trey on September 30, 2009 at 6:47 pm

   Mr. Jackson,
   I agree with you that computing resources need to be closer to the unit and with the concept that the processing and storage can be distributed. Unfortunately, the CIO/G6 and NETCOM strategic vision is to move them farther away and consolidate services in "Army Processing Centers" which will not be in theater at all. At this time they are still planning on equipping the APCs with conventional server farms instead of the virtual server architecture advocated by DISA's RACE pilot. APC White Paper

   I am afraid that the Army is going in two different directions that are technologically inconsistent. I was hoping that your panel might consider this in your defense application discussions.

   MAJ Trey Blacklock
   Information Systems Management Officer
   United States Army

   The views expressed in this BLOG are those of the author and do not reflect the official policy or position of the Department Of the Army, Department of Defense or the U.S. Government

   Reply