Use Case: Portfolio Management
A major benefit of the Evident ClearStone product is its ability to process the disparate data collected and enriched from servers (physical, virtual and grids), data grids, the network and applications into a holistic view of the infrastructure being consumed by virtualized applications. This extends well beyond just a usage and performance view of the resources associated with the application; it also presents a view of the affinity (relationship) of all resources associated with the various infrastructure components that support the application/service. In order to illustrate these benefits, consider the following use case example:
Application Scenario: Portfolio Management
Consider a Portfolio Management Application that contains a component that uses
a Monte Carlo Simulation. The simulation is used to increase accuracy and reliability of financial
planning by calculating the probability of specific outcomes by examining many
different scenarios. By modeling different scenarios the investor avoids
surprises due to risk factors that may negatively impact the results of
portfolios nearing distribution. These factors could include a market downturn
or a higher-than-expected rate of inflation that can affect the return of the
portfolio and the availability of funds due to the timing of fund distribution.
A Monte Carlo Simulation is often used to randomly generate numbers for
uncertain variables such as interest rates, investment volatility, life
expectancy, variable rates of inflation, and other variables such that the
investor can look at different “what if” scenarios
and make decisions with respect to their portfolios based on the odds of
outcomes of different scenarios.
The application environment shown above illustrates a sophisticated implementation of the Portfolio Management Application that uses a heterogeneous compute grid to accomplish many simultaneous Monte Carlo Simulations for portfolio analysis, and uses a data grid to improve the overall application performance by using a distributed cache. The application could be hosting hundreds of end-users, so providing guaranteed response time is paramount to the business owners. The tasks on the grid are also interacting with other enterprise components outside of the grids.
Application Scenario: Application /
Infrastructure Components
The compute and data grid environment illustrated above consist of the following components:
- Portfolio Management Application (PMA) – this is a financial services application that provides various financial
management services to investors including simulations of “what if” scenarios
that include Monte Carlo Simulations.
- Grid Application/Services – these are separate grid-enabled services that perform certain
compute functions on the grid (e.g. a Monte Carlo Simulation).
- Application Data Cache – a distributed cache for data required by the PMA application or by the
calculations being executed in the compute grid.
- External Sources – required interactions with other
financial applications over an Enterprise Service Bus (ESB), external
databases, financial systems, and/or live market data in order to perform risk
analysis related to the application “what if” scenarios.
In this scenario, the Evident ClearStone system is used to measure usage/ consumption and performance of the overall PMA application including the correlated use of the compute grid, the data cache and the portion of the enterprise network that is associated with the compute and data grid and application resources being used by PMA. The collected information, enriched by the Evident ClearStone product and analytics, can provide answers to many questions pertaining to the application and the infrastructure supporting the application.
Actionable Information
The following illustrates the types of questions that can be addressed using the Evident ClearStone reporting information:
| Description |
 |
Questions Addressed via
Evident
ClearStone Reporting |
| |
| Target Audience: PMA
Application Owners |
| |
|
|
| This is the business unit owner of the PMA application that is providing the service to investors. This team is concerned with providing the highest level of service to the end-user. |
|
- How many requests were submitted for “what if” simulations?
- What was the average response time for users who submitted a “what if” simulation?
- How many user-requests were not processed within the criteria imposed by the contracted SLA?
- What is the pattern (over time) of requests for “what if” simulations?
|
| Target Audience: Grid
and/or PMA Application Operational Staff |
| |
|
|
| This is the IT operational staff that is responsible for the day-to-day management and support of the PMA application, the grids and grid services that support the PMA application, and the enterprise network that ties these components together in support of the application. This team is concerned with meeting or exceeding the SLA’s agreed to with the application owners. |
|
- What was the load and activity of “what if” simulations across a specific day, week, or month?
- What was the cache put/get performance for the cache associated with the PMA application?
- What was the relationship of “what if” compute requests and cache activity?
- How does the number of “what if” simulation requests compare to the average response time for a request over time?
- Were there sufficient resources during a “what if” simulation?
- What was the precise amount of time each “what if” simulations spent on the compute grid?
- Was the cache optimal? i.e. partition-size, cache Gets/Puts average, memory utilization, response time.
- What is the affinity (i.e. who is talking to who) among compute grid components and external resources?
- What is the volume of network traffic between different members of a named cache? Which cache members have the largest network transactions and how is that related to members leaving and joining the cache?
- How many and what is the size of the named caches across the data grid?
|
Note that in order to answers some of these questions, the correlation of the data metrics has to be across multiple sources (e.g. combined data from the compute grid and the data grid, or from the data grid and the enterprise network, or from all three sources). This use case illustrates the holistic nature of the Evident ClearStone product and its ability to correlate data from a variety of sources and present a single view of application resources consumption and key performance indicators. This integrated view is important for IT operational staff or IT architects and designers who are looking to analyze virtualized application and infrastructure function and performance.
|