Services and Complexity Roger Sessions CTO ObjectWatch

The Questions Many SOAs fail. Why? What can we do about it?

The project will not meet business needs. The project will have poor quality. The project will be delivered late. The project will exceed its budget. “ At-Risk ” IT Projects

60% of IT budgets are invested in at-risk projects. Costing the world economy 6.5 trillion USD per year Costing the New Zealand economy 21 billion NZD per year The problem is getting worse. Exponentially! Who cares?

What ’ s Causing the Problem? Complexity!

What About Complexity Failure rates are proportional to complexity Cost is proportional to complexity Agility is inversely proportional to complexity Security is inversely proportional to complexity. Maintainability is inversely proportional to complexity. The Bottom Line: Complexity is Very Bad

The Claim Say architecture A and B both solve the business problem Say A is twice as complex as B Then A will cost twice as much, work half as well, and be twice as likely to fail as B

Example: Interlibrary Loan Library 1Library 2 Inter Library Loan Agency

Solution One Borrower ILLA Lender Borrow Book Process Positive Response Process Negative Response Process Borrow Request Request IL Loan Process Positive Lender Response Process Negative Lender Response Respond to Borrower Receive New Book Item Update Catalog Process ILLA Request Agree to Loan Request Reject Loan Request Process New Book

Solution Two Borrower ILLA Lender Borrow Book Process Positive Response Process Negative Response Receive Catalog Receive New Book Item Update Catalog Process Borrow Request Agree to Loan Request Reject Loan Request Process New Book Which SOA is least complex?

- Facts and Fallacies of Software Engineering by Robert Glass For every 25 percent increase in problem complexity, there is a 100 percent increase in solution complexity. Glass’s Law

For every 25 percent increase in the business functionality in a service, there is a 100 percent increase in the complexity of that service. Glass’s Law Applied to SOAs For every 25 percent increase in the number of connections in a service, there is a 100 percent increase in the complexity of that service.

SCU = Standard Complexity Unit The amount of complexity in an average atomic business function implemented in an idealized green field environment with no internal or external interactions. Examples of atomic business functions: process-check, hire-employee, remove- from-inventory

1.25 functionality = 2 X complexity Glass’s Law Fx Cx Bird’s Formula A system with bf business functions is 10 log(Cx) x log(bf) / log(Fx) times more complex than a system of 1 business function. Bird’s Formula

is with constants replaced Equivalent to Simplifying Bird’s Formula 10 log(2) x log(bf) / log(1.25) 10 log(Cx) x log(bf) / log(Fx) x log(bf) / Equivalent to log(bf) Glass’s Constant

A system with bf business functions and cn connections has a complexity value (in SCUs) of 10 log(Cx) x log(bf) / log(Fx) + 10 log(Cx) x log(cn) / log(Fx) Sessions’s Formula for Service Complexity Equivalent to log(bf) log(cn)

An SOA with m services has a total complexity (in SCUs) of Sessions’s Formula for SOA Complexity ∑ i=1 m log(bfi) log(cni)

Complexity of Solution One Borrower ILLA Lender Borrow Book Process Positive Response Process Negative Response Process Borrow Request Request IL Loan Process Positive Lender Response Process Negative Lender Response Respond to Borrower Receive New Book Item Update Catalog Process ILLA Request Agree to Loan Request Reject Loan Request Process New Book bdcompcncomptotal Total: 1009

Complexity of Solution Two Borrower ILLA Lender Borrow Book Process Positive Response Process Negative Response Receive Catalog Receive New Book Item Update Catalog Process Borrow Request Agree to Loan Request Reject Loan Request Process New Book bdcompcncomptotal Total: 314

Two SOA Solutions Solution 1: 1009 SCUs Solution 2: 314 SCUs Solution 1 will cost 3 times more than Solution 2 and be 3 times more likely to fail.

But We Have a Problem We can’t exhaustively examine every possible solution. Why not? For 20 business functions, there are more than 50 trillion SOA solutions. We need a directed process.

SIP Simple Iterative Partitions

The Process - Identification Inter Library Loan Borrow Loan Ask to borrow Process positive response Process negative response Receive catalog Process request Agree to loan Refuse to loan Process new book Add book to catalog

The Process - Partitioning Ask to borrow Process positive response Process negative response Receive catalog Process loan request Agree to loan Refuse to loan Process new book Add book to catalog A A A A B B B B C

The Process - Subsetting Ask to borrow Process positive response Process negative response Receive catalog Process loan request Agree to loan Refuse to loan Process new book Add book to catalog A A A A B B B B C

The Process - Connecting Ask to borrow Process positive response Process negative response Receive catalog Process loan request Agree to loan Refuse to loan Process new book Add book to catalog

Goals of SIP A process to drive the simplest possible SOA that solves a given business problem Based on mathematical models of complexity, partitioning, and set theory. Reproducible and verifiable.

Summary IT is in a multi-trillion dollar crisis. The major cause is complexity We must build the simplest possible SOAs. SIP is a process for finding the simplest possible SOA.

For more information

Sessions On-Demand & Community Resources for IT Professionals Resources for Developers Microsoft Certification & Training Resources Resources

COMPLETE YOUR EVALUATION FORMS IN COMMNET AND BE IN TO WIN ONE OF THE 150 DAILY PRIZES* GIVE US YOUR FEEDBACK & WIN INSTANTLY! *For full terms & conditions and more information, please visit the CommNet Portal.

© 2009 Microsoft Corporation. All rights reserved. Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.