How to rebuild a manufacturing travel card for mobile age, from Crystal Report to Rails 4 web application.

1. 1
Web TCard – Speed optimization
Eric Guo
MES Business Systems Analyst
3/22/2015

2. 2
Overview
 Background & Objective
 Experimental - Measure & Finding bottleneck
 Result & Discussion
 Conclusion
 Learnt from it

3. 3
Background & Objective
Speed
More data source.
Stable (0 downtime)
Agility
Built for mobile instead of paper.
Built for next 10 years.

4. 4
Measure & Finding bottleneck in CR report
 Using M1512ME9F3.02 and MT TCard as example.
 http://cvpmesip04/TCardPrint/
– Total loading time 3.86s
• Including 16 sub report, time in ms
• 103+49+228+71+240+46+55+110+49+55+89+49+18+47=1209 MS
• All rest time is relative with render or running in Crystal Report
internally.

5. 5
Latency times every programmer should know
L1 cache reference 0.5 ns
Branch mispredict 5 ns
L2 cache reference 7 ns 14x L1 cache
Mutex lock/unlock 25 ns
Main memory reference 100 ns 4x mutex op, 20x L2 cache, 200x L1 cache
Compress 1K bytes with Zippy 3,000 ns
Send 2K bytes over 1 Gbps network 20,000 ns
Read 1 MB sequentially from memory 250,000 ns
Round trip within same datacenter 500,000 ns
Disk seek 10,000,000 ns 20x datacenter roundtrip
Read 1 MB sequentially from disk 20,000,000 ns 80x reading in seq from
memory
Send packet CA->Netherlands->CA 150,000,000 ns
By Jeff Dean (http://research.google.com/people/jeff/):

6. 6
Analysis and feasibility analyze
 Sub-report slowdown the total processing
– But it’s needed and necessary because of biz requirement
 Deep link table join will cause slowness
– But you need link those table to take data to be showing in
T-Card latter.
 Crystal Report takes 2.6+ seconds to do internal works
– But it’s almost can not make any change to affect it
 Crystal Report will failed to loading report sometime
– No source code, no way to improve.

7. 7
Days of future past – act like X-Men
 Sub-report slowdown the total
processing
– Stop using sub-report in new web t-
card
 Deep link table join will cause
slowness
– Stop using any join when query DB
table
 Crystal Report takes 2.6+ seconds to
do internal works
– No internal code, everything should
can be changed
 Crystal Report will failed to loading
report sometime
– Stop using Crystal Report

8. 8
Perception is Reality
 Stop using sub-report in new web t-card
– Fetch data in one place
 Stop using any join when query DB table
– One query per one table, in-memory calculation all T-Card
info, do filter, join.
 No internal code, everything should can be changed
– Know each line of time taken
 Stop using Crystal Report
– Whole web T-Card built on the open source stack

9. 9
Trace the SQL payload by line

10. 10
Result & Discussion
 M1512ME9F3.02 is now takes 0.68s to loading
 5 times faster.

11. 11
Conclusion & Learnt From It
 Make things faster probably need re-think the whole
design.
 Except make things faster, many things also important.
 Maintain cost
 Support cost
 Stability
 Simplicity
 The more you known, the more freedom you have to
balanced contradiction requirement
– http://cvpscmip01/mv.git