Using available resources efficiently is not easy: our designs are limited by market
domination of existing systems, technological limitations and created sometimes
artificially demand of, frankly speaking, completely non-necessary applications. At
the same time, we have to ask ourselves:
• What else is possible to add in our systems? and
• How to address real problems to assist human beings and living using
technologies?
Considering that DCS consists of over 3 billion computers and dozens of billions of
controllers, all connected, the challenge in handling this world for good is astronomical. Surprisingly, some hints how to address these challenges were already a
subject of R&D efforts of scientists from ETH Zurich and IT-ACS Ltd, their papers
[1, 2], books [3–5], interviews [6], and keynote speeches (the latest is: [7]). In short,
we can summarize challenges as combination of essential non-functional requirements (NFR) for the DCS and stand-alone future CS in terms of as Performance-,
Reliability- and Energy-efficiency.
Thus, widening application area for ICT (information and computer technology)
force us to think about computers, distributed computer systems and structures from
a perspective of resources, available redundancy, then see their applications in
terms of improvements and efficiency of a system as declared in [1–7], considering
three strongly dependent requirements:
• performance
• reliability
• energy efficiency (power consumption)
When the system is considered to be used by various applications which have
various demands and these demands also vary from application to application we
simply must think about design of the system with capabilities of trading available
resources for different purposes. Trading resources here means that, for example, if
applications requested maximum performance then the system should use its
resources for that and even borrow from resources previously allocated for
reliability.
There is no doubt that implementation of these non-functional requirements
needs some redundancy in the software and hardware [3–5]. Note that redundancy
is already a property, and its efficient use is a goal to satisfy mentioned NFR of ICT,
primarily in DCS domain. It is worth to mention here also, that handling redundancy for various purposes, addressing various and changeable requirements
becomes actually a process.
Since late 60’s up to early 80’s of last century researchers were questioned
themselves on how to design computers systems for the future, making them
reliable, resilient, power efficient and powerful. 50 years on and we still have the
same questions. The discussion now is more focused on properties and features
required for next generation of computer systems, especially distributed computer
systems (DCS), that includes networking and clustered computers.
250
18 Distributed Systems: Maximizing Resilience
domination of existing systems, technological limitations and created sometimes
artificially demand of, frankly speaking, completely non-necessary applications. At
the same time, we have to ask ourselves:
• What else is possible to add in our systems? and
• How to address real problems to assist human beings and living using
technologies?
Considering that DCS consists of over 3 billion computers and dozens of billions of
controllers, all connected, the challenge in handling this world for good is astronomical. Surprisingly, some hints how to address these challenges were already a
subject of R&D efforts of scientists from ETH Zurich and IT-ACS Ltd, their papers
[1, 2], books [3–5], interviews [6], and keynote speeches (the latest is: [7]). In short,
we can summarize challenges as combination of essential non-functional requirements (NFR) for the DCS and stand-alone future CS in terms of as Performance-,
Reliability- and Energy-efficiency.
Thus, widening application area for ICT (information and computer technology)
force us to think about computers, distributed computer systems and structures from
a perspective of resources, available redundancy, then see their applications in
terms of improvements and efficiency of a system as declared in [1–7], considering
three strongly dependent requirements:
• performance
• reliability
• energy efficiency (power consumption)
When the system is considered to be used by various applications which have
various demands and these demands also vary from application to application we
simply must think about design of the system with capabilities of trading available
resources for different purposes. Trading resources here means that, for example, if
applications requested maximum performance then the system should use its
resources for that and even borrow from resources previously allocated for
reliability.
There is no doubt that implementation of these non-functional requirements
needs some redundancy in the software and hardware [3–5]. Note that redundancy
is already a property, and its efficient use is a goal to satisfy mentioned NFR of ICT,
primarily in DCS domain. It is worth to mention here also, that handling redundancy for various purposes, addressing various and changeable requirements
becomes actually a process.
Since late 60’s up to early 80’s of last century researchers were questioned
themselves on how to design computers systems for the future, making them
reliable, resilient, power efficient and powerful. 50 years on and we still have the
same questions. The discussion now is more focused on properties and features
required for next generation of computer systems, especially distributed computer
systems (DCS), that includes networking and clustered computers.
250
18 Distributed Systems: Maximizing Resilience