Concerns and issues relating to all versions of WebSphere Application Server
Recently I got a chance to review the "IBM WebSphere eXtreme Scale 6
" book from PACKT Publishing by Anthony Chavez.
Overall the book is well written and a good start to the world of In Memory Data Grids (IMDGs) particularly WebSphere eXtreme Scale (WXS). Before diving into the book I recommend reading some introductory material on ORM Frameworks and Dependency Injection. The first seven chapters of the book are of easy/intermediate complexity. The last three chapters dealing with "Data Grid Patterns", "Spring Integration" and "Putting It All Together" are difficult to grasp without some the background information alluded too earlier. These chapters are also the most valuable, as they help a developer understand how to retrofit WebSphere eXtreme Scale to existing applications.
This book does NOT cover the features introduced in WXS 7 and value-add features when WXS is integrated with the WebSphere Application Server. (like Dynacache and HTTPSession integration)
The broad concepts of high scalability explained in the book are applicable to any of the IMDGs out there. Therefore I request any developer interested in these new memory architectures to read this book.
Below is a review of each chapter
Chapter 1: What is a Data Grid
Good Explanatation on the need for datagrids for scalability and Data Grid Basics. A little bit of information about the competitors of WXS in this space would have been beneficial to the user (See Relevant link #2).
Chapter 2: ObjectMap API
The explanation of Lock types is a bit laborious and can be simplified. Optimistic and Pessimistic lock strategies have been explained with good examples.
Chapter 3 Entities & Queries
Reading this chapter it was unclear on when to use JPA entity anotations and when to use eXtreme Scale entity annotations. The relationship between ORM-JPA and WebSphere eXtreme Scale persistence needs to be clarified. (See eXtreme Scale podcast episode #3 - Relevant link # 3)
Chapter 4: Database Integration
Good explanation of read-through, write-through and write-behind loaders. Difference between JPAEntityLoaders and JPALoaders needs to be made clear with guidance on which loader to use based on application context.
Chapter 5: Handling Increased Load
Well written. Would have liked to see a bigger section on capacity planning with formulas and tools to determine the right size of the eXtreme Scale topology.
Chapter 6: Keeping Data Available:
Some of the concepts explained in this chapter have been repeated.Would have liked to see a much crisper and shorter explanation of partitions, shards & mapsets acros chapters 5 & 6. The sections on controlled data placement and preferred zones are an easy-read.
Chapter 7: The DataGrid API
The origins of map-reduce from functional programming provide good context to the MapAgent and ReduceAgent content that follows. Contents of this chapter are dense however very important because this chapter explains how to colocate code with data by moving the code to the data in the grid.
Chapter 8: Data Grid Patterns
I would liked to see more information on Constrained Tree Schemas (See Relevant link #5.). Concepts of reference data and object duplication, ascertaining the schema root for a data model and object schema denormalization are critical in working with your data model with WXS. Good big picture comparison of In-memory databases and IMDGs in the XTP section.
Chapter 9: Spring Integration
Comprehensive and will most make sense to developers who are familiar with Spring.
Chapter 10: Putting It All Together:
I would have liked to see a more detailed explanation of how the Bookmarks application is architected, particularly its data model. This chapter moves head-on into refactoring an older hibernate application to use eXtreme Scale. Grid/ORM hybrid section is difficult to grok. Caching more than ORM section brings forward an imporant concept that we are not limited to using eXtreme Scale with ORM'd objects.