stasis-aries-wal/doc/position-paper/LLADD.txt

Russell Sears
Eric Brewer
UC Berkeley

A Flexible, Extensible Transaction Framework

Existing transactional systems are designed to handle specific
workloads.  Unfortunately, the implementations of these systems are
monolithic and hide their transactional infrastructures underneath a
SQL interface. Lower-level implementations such as Berkeley DB
efficiently serve a wider variety of workloads and are built in a more
modular fashion.  However, they do not provide APIs to allow
applications to build upon and modify low-level policies such as
allocation strategies, page layout or details of recovery semantics.
Furthermore, data structure implementations are typically not broken
into separable, public APIs, which discourages the implementation of
new transactional data structures.

Modern object-oriented programming languages such as C++ or Java
handle the problem differently.  Such languages typically provide a
large number of data storage algorithm implementations.  These
structures may be used interchangeably with application-specific data
collections, and collection implementations may be composed into more
sophisticated data structures.

We have implemented LLADD (/yad/), an extensible transactional storage
library that takes a composable, layered approach to transactional
storage.  Below, we present some of its high level features and
performance characteristics and discuss our plans to extend the system
into distributed domains.  Finally we introduce our current research
focus, the application of automated program verification and
optimization techniques to application specific extensions.  Such
techniques should significantly enhance the usability and performance
of our system, allowing application developers to implement
sophisticated cross-layer optimizations easily.

Overview of the LLADD Architecture

General-purpose transactional storage systems are extremely complex
and only handle specific types of workloads efficiently.  However, new
types of applications and workloads are introduced on a regular basis.
This results in the implementation of specialized, ad-hoc data storage
systems from scratch, wasting resources and preventing code reuse.

Instead of developing a set of general purpose data structures that
attempt to perform well across many workloads, we have implemented a
lower-level API that makes it easy for application designers to
implement specialized data structures.  Essentially, we have
implemented an extensible navigational database system.  We believe
that this system will support modern development practices and allow
transactions to be used in a wider range of applications.

In order to support our layered data structure implementations and to
support applications that require specialized recovery semantics,
LLADD provides the following functionality:

 - Flexible Page Layouts for low-level control over transactional data
   representations

 - Extensible Log Formats for high level control over transactional
   data structures

 - High and low level control over the log, such as calls to "log this
   operation" or "write a compensation record"

 - In-memory logical logging for data store independent lists of
   application requests, allowing "in flight" log reordering,
   manipulation and durability primitives to be developed

 - Extensible locking API for registration of custom lock managers and
   a generic lock manager implementation

 - Custom durability operations such as save points and two-phase
   commit's prepare call.

We have shown that these primitives allow application developers to
control on-disk data representation, data structure implementations,
the granularity of concurrency, the precise semantics of atomicity,
isolation and durability, request scheduling policies, and deadlock /
avoidance schemes.  The ability to control or replace these modules
and policies allows application developers to leverage application and
workload specific properties to enhance performance.

While implementations of general-purpose systems often lag behind the
requirements of rapidly evolving applications, we believe that the
flexibility of our architecture allows us to address such applications
rapidly.  If the applications in question represent a large enough
market or an important class of workloads, high-level declarative
systems may eventually replace lower level approaches based on our
system.  For applications that represent a small market, the
implementation of such high-level declarative systems is probably not
feasible.  In these cases, our system could provide a reasonable
long-term solution.

For example, XML storage systems are rapidly evolving but still fail
to handle many types of applications.  Typical bioinformatics data
sets [PDB, NCBI, GO] must be processed by computationally intensive
applications with rigid data layout requirements.  The maintainers of
these systems are slowly transitioning to XML, which is valuable as an
interchange format and is also supported by many general purpose
tools.  However, many of the data processing applications that use
these databases still must employ ad-hoc solutions for computationally
expensive tasks and data production pipelines.

XML database systems may eventually meet all of the needs of of these
scientific applications.  However, extensions implemented on top of a
flexible storage system could have avoided the need for ad-hoc
solutions, and served as a prototype for components of higher level
implementations.

LLADD is based upon an extensible version of ARIES [ARIES] but does
not hard-code details such as page format or data structure
implementation.  It provides a number of "operation" implementations
which consist of redo/undo methods and wrapper functions.  The
redo/undo methods manipulate the page file by applying log entries
while the wrapper functions produce log entries.  Redo methods handle
all page file manipulation during normal forward operation, reducing
the amount of code required to implement new data structures.  LLADD
handles the scheduling of redo/undo invocations, disk I/O, and all of
the other details specified by the ARIES recovery algorithm. This
allows operation implementors to focus on the details that are
important to the functionality their extension provides.

LLADD ships with a number of default data structures and layouts,
ranging from byte-level page layouts to linear hashtables and
application-specific recovery schemes and data structures.  These
structures were developed with reusability in mind, encouraging
developers to compose existing operations into application-specific
data structures.  For example, the hashtable is implemented on top of
reusable modules that implement a resizable array and two exchangeable
linked-list variants.

In other work, we show that the system is competitive with Berkeley DB
on traditional (hashtable based) workloads, and show significant
performance improvements for less conventional workloads including
custom data structure implementations, graph traversal algorithms and
transactional object persistence workloads.

The transactional object persistence system is based upon the
observation that most object persistence schemes cache a second copy
of each in-memory object in a page file, and often keep a third copy
in operating system cache.  By implementing custom operations that
assume the program maintains a correctly implemented object cache, we
allow LLADD to service object update requests without updating the
page file.

Since LLADD implements no-force, the only reason to update the page
file is to service future application read requests.  Therefore, we
defer page file updates until the object is evicted from the
application's object cache. This eliminates the need to maintain a
large page cache in order to efficiently service write requests.  We
also leveraged our customizable log format to log differences to
objects instead of entire copies of objects.

With these optimizations, we showed a 2-3x performance improvement
over Berkeley DB on object persistence across our benchmarks, and a
3-4x improvement over an in-process version of MySQL with the InnoDB
backend.  (A traditional MySQL setup that made use of a separate
server process was prohibitively slow.  InnoDB provided the best
performance among MySQL's durable storage managers.)  Furthermore, our
system uses memory more efficiently, increasing its performance
advantage in situations where the size of system memory is a
bottleneck.

We leave systematic performance tuning of LLADD to future work, and
believe that further optimizations will improve performance on these
benchmarks significantly.

Because of its natural integration into standard system software
development practices, we think that LLADD can be naturally extended
into networked and distributed domains.  For example, typical
write-ahead-logging protocols implicitly implement machine
independent, reorderable log entries in order to implement logical
undo.  These two properties have been crucial in past system software
designs, including data replication, distribution and conflict
resolution algorithms.  Therefore, we plan to provide a networked,
logical redo log as an application-level primitive, and to explore
system designs that leverage this approach.

Current Research Focus

LLADD's design assumes that application developers will implement
high-performance transactional data structures.  However, these data
structures are notoriously difficult to implement correctly.  Our
current research attempts to address these concerns.

For our infrastructure to be generally useful the functionality that
it provides should be efficient, reliable and applicable to new
application domains.  We believe that improvements to the development
process can address each of these goals.

Application developers typically have a limited amount of time to
spend implementing and verifying application-specific storage
extensions, but bugs in these extensions have dire consequences.
Also, while data structure algorithms tend to be simple and easily
understood, performance tuning and verification of implementation
correctness is extremely difficult.

Recovery-based algorithms must behave correctly during forward
operation and also under arbitrary recovery scenarios.  Behavior
during recovery is particularly difficult to verify due to the large
number of materialized page file states that could occur after a
crash.  Fortunately, write-ahead-logging schemes such as ARIES make
use of nested-top-actions to simplify the problem.  Given the
correctness of page-based physical undo and redo, logical undo may
assume that page spanning operations are applied to the data store
atomically.

Existing work in the static-analysis community has verified that
device driver implementations correctly adhere to complex operating
system kernel locking schemes [SLAM]. We would like to formalize
LLADD's latching and logging APIs so that these analyses will be
directly applicable to LLADD.  This would allow us to verify that data
structure behavior during recovery is equivalent to the behavior that
would result if an abort() was issued on each prefix of the log.

By using coarse latches that are held throughout entire logical
operation invocations, we can drastically reduce the size of this
space.  This would allow conventional state-state based search
techniques (such as randomized or exhaustive state-space searches, or
unit testing techniques) to be practical.  It has been shown that such
coarse-grained latching can yield high-performance concurrent data
structures if semantics-preserving optimizations such as page
prefetching are applied [ARIES/IM].

A separate approach to the static analysis of LLADD extensions uses
compiler optimization techniques.  Software built on top of layered
APIs frequently makes calls to low level functions that result in
repeated work.  A common example in LLADD involves loops over data
with good locality in the page file.  The vast majority of the time,
these loops result in a series of high level API calls that
continually pin and unpin the same underlying data.

The code for each of these high level API calls could be copied into
many different variants with different pinning/unpinning and
latching/unlatching behavior.  This would greatly complicate the API
that application developers must work with and complicate any
application code that made use of such optimizations.

Compiler optimization techniques such as code hoisting and partial
common subexpression elimination solve analogous problems to remove
redundant computations.  Code hoisting moves code outside of loops and
conditionals, while partial common subexpression elimination inserts
checks that decide at runtime whether a particular computation is
redundant.  We hope to extend such techniques to reduce the number of
buffer manager and locking calls made by existing code.  In situations
where memory is abundant these calls are a significant performance
bottleneck, especially for read-only operations.

Similar optimization techniques are applicable to application code.
Local LLADD calls are normal function calls.  Therefore it may be
possible to apply the transformations that these optimizations perform
to application code that is unaware of the underlying storage
implementation.  This class of optimizations would be very difficult
to implement with existing transactional storage systems but should
significantly improve application performance.

We hope to validate our ideas about static analysis by incorporating
them into our development process as we increase the reliability and
overall quality of LLADD's implementation and its APIs.

Our architecture provides a set of tools that allow applications to
implement custom transactional data structures and page layouts.  This
avoids "impedance mismatch," simplifying applications and providing
appropriate applications with performance that is comparable or
superior to other general-purpose solutions.  By adding support for
automated code verification and transformations we hope to make it
easy to produce correct extensions and to allow simple, maintainable
implementations to compete with special purpose, hand-optimized code.

Conclusion

We have described a simple, extensible architecture for transactional
systems and presented a number of situations where our implementation
outperforms existing transactional systems.  Due to the flexibility of
the architecture, we believe that it is appropriate for evolving
applications and for applications where general-purpose, declarative
systems are inappropriate.  Finally, we presented a number of
optimizations that our system can support, but that would be extremely
difficult to apply to existing transactional data stores.  Therefore,
we believe that our approach is applicable to a wider range of
scenarios than existing systems.

Acknowledgements

Mike Demmer was responsible for LLADD's object persistence
functionality.  Jimmy Kittiyachavalit, Jim Blomo and Jason Bayer
implemented the original version of LLADD.  Gilad Arnold, and Amir
Kamil provided invaluable feedback regarding LLADD's API.

[SLAM]     Ball, Thomas and Rajamani, Sriram.  "Automatically 
	   Validating Temporal Safety Properties of Interfaces," 
	   International Workshop on SPIN Model Checking, 2001.
[GO]       Gene Ontology, http://www.geneontology.org/
[ARIES]    C. Mohan, Don Haderle, Bruce Lindsay, Hamid Pirahesh, 
	   Peter Schwarz.  "ARIES: a transaction recovery method 
	   supporting fine-granularity locking and partial 
	   rollbacks using write-ahead logging," TODS, 1992.
[ARIES/IM] C. Mohan, Frank Levine.  "ARIES/IM: an efficient and 
	   high concurrency index management method using 
	   write-ahead logging," ACM SIGMOD, 1992.
[NCBI]	   National Center for Biotechnology Information, 
	   http://www.ncbi.nlm.nih.gov/
[PDB]      Protein Data Bank, http://www.rcsb.org/pdb/