White Papers & Analyst Reports
 |
Summary of
Database Performance Comparison
By KLAS Enterprises - January 2004
In
January 2004, KLAS Enterprises published a report entitled “Database
Performance:
A Comparison of Applications Based on Caché- and Oracle-Based
Applications”. The aim
of this study was to collect real-world information about the
performance of leading
database management systems in large-scale healthcare applications.
In summary:
“There is sufficient independent data to validate a significant variance between Caché and Oracle in several significant areas, especially with respect to the care and feeding of the database system. Overall, Caché leads in all five of the specific performance measurements. The following area stands out in the clinical area:”
- InterSystems' database products outperformed Oracle's by more than 1.3:1 in hardware required and 1.4:1 in DBAs required.
How the Data Was Collected
KLAS is a research and consulting firm that specializes in monitoring the performance of healthcare IT vendors. It regularly surveys executives and clinicians from more than 4,500 healthcare facilities to get their real-world perspective on 500+ different products from 300+ vendors.
For this study, KLAS contacted 100 organizations that use four well-known applications, two based on InterSystems' database products and two based on Oracle's database products. For an enterprise-wide perspective, KLAS surveyed users of two electronic medical records applications. For a departmental perspective, KLAS surveyed users of two laboratory applications.
Respondents were asked a combination of satisfaction questions (e.g., "On a scale of 1 to 9, how satisfied are you with the reliability of the database used for this application?") as well as more specific quantitative questions (e.g., "How many DBAs do you have?")
WWhat Users Reported about End User & Administrative Response Times
When looking at the overall results we note that Caché leads in all five of the specific performance measurements. In the performance rating areas, those revolving around uptime, response time, reliability, and speed and scalability, Oracle also lagged behind Caché in all areas. These areas were all measured on a scale of 1-9, with 9 being high.
- The biggest differentiators were uptime and end-user/administrative response time.
- Both databases/applications had the lowest satisfaction
score in the same area:
administrative response time. (Leading area denoted with a
).
|
Criterion |
Caché |
Oracle |
|
Response times by database type: |
|
|
end user |
7.93 |
7.24 |
administrative (scale of 1 - 9) |
7.93 |
7.19 |
Criterion |
Caché |
Oracle |
Satisfaction with the reliability of the database system for this application (scale of 1 - 9) |
8.30 |
8.18 |
What Users Reported about Performance and Scalability
KLAS asked users to rate their satisfaction with the speed and scalability of the database used for their healthcare applications. Caché scored higher.
Criterion |
Caché |
Oracle |
Satisfaction with speed and scalability of |
7.95 |
7.63 |
The true measure of a database is its performance in live operation. As we continue to measure and collect real data, we are able to cut through all the hype about technology, state-of-the- art solutions, object oriented systems, relational databases, and client/server. In doing so, we continue to see that Caché serves the healthcare community as a core component in the best-rated systems in the market. This also holds true with the uptime of the system.
Criterion |
Caché |
Oracle |
Satisfaction with uptime (scale of 1 - 9) |
8.29 |
7.82 |
User Reported: Length of Longest Unscheduled Downtime
|
Criterion |
Caché |
Oracle |
|
Duration of longest unscheduled downtime |
|
|
1-2 hrs |
24% |
12% |
3-6 hrs |
33% |
16% |
7-12 hrs |
17% |
23% |
13-24 hrs |
13% |
42% |
User Reported: Number of Scheduled Downtimes in the Past Year
Criterion |
Caché |
Oracle |
Duration of longest unscheduled downtime |
|
|
None |
2% |
2% |
1 |
13% |
4% |
2-3 |
42% |
13% |
4-6 |
19% |
30% |
Over 6 |
25% |
50% |
What Users Reported on Database Adminstrators (DBAs) Needed to Manage the Database
Criterion |
Caché |
Oracle |
Duration of longest unscheduled downtime |
|
|
Average number of DBAs |
1.13% |
1.6% |
Bottom Line
KLAS summarized the results as follows:
“When looking at the overall results (all Caché-based products vs. all Oracle-based products) we note that Caché leads in all five of the specific performance measurements. The largest variation was found between the two databases in administrative response times”.