How to Design for the Cloud

Unlike traditional methods, cloud platforms and cloud services produce rapid solution delivery, a massive array of flexible elements within a solution and/or cross-functional data collection. With innovative technologies, cloud providers can produce sophisticated dashboards, configurability and monitoring to their clients.

Although cloud technologies have advanced considerably, modifications in the client application design still stand room for improvement. The concepts of application architecture, as was referenced at the WSTA Cloud Technologies seminar in 2011, will inevitably include parallel computing, state machines, mutexes and highly accessible programming processes.

Parallel Processing

Parallel processing design is a necessity for many time-sensitive and high volume applications. There is significant uncertainty, however, of how many applications would handle the dynamic assignment of multiple instances or would perform in a cloud-based environment. Many mission-critical systems, such as legacy applications, are not migrated to the cloud based on these factors. Therefore, the migration of applications to a cloud-based atmosphere requires expertise on the overall operations of an application and not just its functional logic to be performed effectively.

Multi-Node (Inter-Process) Messaging

Message complexity for many solutions in the financial industry come in several forms :

  • Asynchronicity – Asynchronicity involves the managing broadcast message requests for diverse data across the Internet from various timeframes into a view, package or product that is uniform in nature and accurate.
  • Modeling – Modeling is done to help a company determine how to reach a certain objective the most favorably through the application of many filters, settings and strategies to data products.

Utilization of these features can be done through the design and use of parallel processing and message synchronization. Although cloud service providers are capable of providing an enormous parallel architecture, a properly constructed and modular software design is required first in order to support parallelism.

Options

The use of a strong inter-process messaging system is key when designing software to become cloud-friendly and includes the implementation of two critical features:

  • Store-and-Forward: This describes a message’s capability of being indefinitely held or timed-out (by a few rules) while still being accessible if it ever needs to be retrieved by a receiving system. Unlike the many-to-one messaging paradigm that exists in traditional solutions like Java Message Service (JMS) and MQ, messaging queues and listeners must exist in a many-to-many ratio in order to be cloud-friendly.
  • Multi-Route: Involves the maintenance of switching routes between primary nodes via the most efficient and least expensive method possible. The multi-route capability already exists in many border gateway protocol (BGP) routers, and requires logic that is more complex than just resource monitoring and heartbeats to be implemented to a many-to-many cloud infrastructure.

Complexity

The utilization of Store Forward and Multi-route are dependent on multiplicity and singularity; these are just two examples of messaging strategies needed by the application.

Multiplicity

Multiplicity is a messaging strategy that must do several things when executing very large identical components, and should ideally have a positive end result:

  • Multiple instances of a component need to run against the same data elements efficiently.
  • Multiple nodes in the cloud must process the same message in parallel when a queued message exists to speed up processing.

Singularity

Singularity confirms the message is transmitted and executed only one time; execution may involve a critical path or permanent information changes.

A combination of tactics may need to be used for an application depending on the criteria of communication in a system between individual modules. Also, the message needs to target a strategy’s features, such as relationship and work capacity.

Considerations

Infrastructure lock-in continues to be the main hurdle of designing a complex application to a strong cloud-based inter-process messaging system. Therefore, a universal standard messaging system that needs to be implemented for all rivaling cloud providers. Designing RAID-like failover to an application includes the following:

1. Find a compatible approach to the Agile SDLC process by modularizing functionality.

2. Design an application communication that is multifunctional. It needs to have an available parallel execution, synchronization and message queuing system.

Protection and privacy of application data is imperative before any applications or data is migrated to the cloud and proper attention must be made to proper key management procedures, such as with data encryption appliances, transparent database encryption and with application-level encryption.

 

Victor Daily is a tech blogger with years of experience in sharing useful tips and fresh news with his readers. His articles have been published on numerous industry blogs. Valuable info for this article has kindly been provided by VPS Australia.

Related Posts:

  • No Related Posts

1 Response to "How to Design for the Cloud"

  • brett beaubouef 12:26 PM 12/7/2013

    SaaS ERP is not a push-button solution. SaaS ERP is the latest effort in the ERP industry to provide a rapid, cost-effective solution for customers who want an enterprise solution. A SaaS deployment model does provide the potential for greater value realization; however, the value proposition is dependent upon appropriate expectations and implementation approach. The purpose of the following article is to provide insight to ensure customers make realistic and informed decisions.

    http://gbeaubouef.wordpress.com/2013/07/10/saas-erp-is-not-push-button/

Leave a Reply

Your email address will not be published. Required fields are marked *