Multicellular Computing: Parallels between Biology and Computing
Biological systems are complex, evolving, information processing systems that are driven toward multicellularity as a way to manage ever-growing complexity. For similar reasons, computing is now driven inexorably toward multicellular architectures.
As computing becomes an ever more pervasive part of our modern lives, we
naturally look to understand computing in terms of what we know of other
phenomena such as biology. And vice versa. That is, each can be a useful
metaphor for understanding the other. On the one hand, biological systems show
computing professionals the way to the exceedingly sophisticated kinds of collaboration
found in bilogical organisms between cells, between tissues, between organs and
even between organisms collaborating/competing in an ecology. On the other hand,
computing systems are recent and are human designed. We can still remember the
stages of evolution in computing and thereby perhaps get insight into design
constraints in the evolution of computing that may have had parallels
in biology that have been lost in the mists of prehistory.
Parallels between biology and computing systems include:
- They are information processing systems.
- Both are examples of highly
ordered yet incredibly complex systems.
- Both evolve over time.
- They both evolved hierarchical encapsulation schemes to control unwanted interactions.
Biology
The simpler Metazoans (i.e., multicellular organsims) include:
Placozoa -- Thought to be the simplest Metazoan organism.
They are millimeter-scale discs that contain a few thousand cells.
The cell types are: Dorsal cells with flagella, ventral “gland cells,” ventral flagellar cells, and central “fiber cells ” that have some of the properties of both nerve cells and muscle cells.
Hydra -- A milimeter-scale (5-20mm), multi-tentacled Cnidaria (the family that includes jellyfish).
It is estimated to contain 50,000 - 70,000 cells. It has at least 15 cell types including simple nerves, epidermis (outer layer), gastrodermis (inner layer), germ cells, and Nematocysts (stinging cells).
Jellyfish (e.g., Cnideria Cyanea) -- Number of cells >
10 million. At least 22 cell types including neurons, sensors, muscle,
endocrine, and Nematocysts used for capturing prey.
Computing
There are many examples. For example, Wikipedia, large public database services (e.g., EMBL/EBI Hinxton GenBank), various "clouds" (Google's Cloud, Amazon's EC2), Social networks (Facebook, Twitter), Massively Multiplayer Online Role Playing Games (World of Warcraft, EverQuest, Second Life) and instant messaging, chat and VOIP systems (e.g., Skype). The largest and most familiar, and those about which more data is available, are:
Ebay -- Estimated 15K-20K servers, At least 5 functional types of servers: database,
LDAP, web servers, application servers, networking switches and routers
Facebook -- An estimated 30,000 servers as of the end of 2009. There are at least 4 types: proxy servers, web servers, database servers, and application servers.
Google -- uses at least 1.5 million servers and very likely more than 2 million servers.
Their system uses at least 9 specialized types of server: crawers, ad servers, indexing, spelling, documents, http, search, formatting, and proxy/cache.
Note that Facebook and EBay, while among the largest multicellular computing systems, are comparable to Placozoa, the smallest Metazoan. And Google, by far the largest multicellular computing system is far simpler than a jellyfish. Moreover, a jellyfish is self organizing, self sustaining, and self reproducing whereas Google is none of those things.
Neither biological nor computing systems advance just by growing more and more "cells" of one type. They seem to prefer to specialize and exploit more types. The chart below has far too few data points to draw much of a conclusion, yet it suggests a possible relationship between the number of "cells" and the number of "cell" types in both realms.
Contact: sburbeck at mindspring.com
Last revised 5/24/2010