Server located in Toronto
Internet Speed 38 Mb Down, 6 Mb Up
User 1 Local
Workstation
Internet Speed 38 Mb Down, 6 Mb Up
Connected to server via WAN
User 2 in
Oshawa (50 Km away)
Internet Speed 6Mb Down, 0.75 Mb Up
Both users are logged into the server during the test. Record locks are respected. The application has 48 tables and 35 indexes and opens all of them at startup.
Test results (All times are in seconds)
Function User 2 User 1
Open all
files (40 DBF, 35 CDX, 80 tags) 15 1
Browse paging (1 Page - 12 lines per page) 6 <1
Simple report (list, 1 page) 4 <1
Report with data selection ½ page 35 <1
Report with data selection 7 pages 540 3
Customer balances report (4 pages) 70 1
User 2 has reported that the application is too slow for his liking.
These test results are from an application written back in 1986 and is now being converted to Harbour without significant code changes. Is there a way to improve the WAN performance?
Regards.
Ash
Test results (All times are in seconds)
Function User 2 User 1
Open all files (40 DBF, 35 CDX, 80 tags) 15 1
Browse paging (1 Page - 12 lines per page) 6 <1
Simple report (list, 1 page) 4 <1Report with data selection ½ page 35 <1
Report with data selection 7 pages 540 3
Customer balances report (4 pages) 70 1
User 2 has reported that the application is too slow for his liking.
These test results are from an application written back in 1986 and is now being converted to Harbour without significant code changes. Is there a way to improve the WAN performance?
Hello Ash,
the 5.th param of netio_Connect() is for compression level.
Not yet myself checked, as i need best speed in LAN, but maybe try:
HB_ZLIB_COMPRESSION_SPEED
As you certainly know, compress and decompress needs CPU power ...
And i'm unsure if this compression applies for all network action of HBNETIO.
Other values in hbzlib.ch
Please report if that changes something ..
best regards
Rolf
User 2 was too generic... what is slow ? Everything ? Only the reports?
Anyway, in a similar situation I use a VPN to log into the intranet where the server is and use remote desktop on a workstation that is local to the server.
As you said, the app is from 1986 and probably has some code that may be changed to use quicker algos.
Recently I sould see some very old code and it had several redundant code... code that was really quick when run once but made the program crawl when in a loop... or, in your case, when accessing the record takes long times over a wan
Hi Ash,
hmmm ...
just checked, to get a connection with enabled 'compression' lead to no problem for me ..
.. but i am very quite unsure if compression is possible for work with DBF and not only for so called DataStreams ...
Because exact there i see the bottleneck for you:
6 Mega- BIT or BYTE ? internet connection, especially with an upload of only 0.75 x:
that is no question, very sure, *slowly* for database work over internet for your client 2 ..
best regards
Rolf
Tried the following code but the connection fails.
Anyway, in a similar situation I use a VPN to log into the intranet where the server is and use remote desktop on a workstation that is local to the server.
That would allow one RDP user, correct? I need a number of users accessing the server at the same time. Hence my move to NetIO.
Not necessarily. For example, TSPlus and WinFlector allow multiple connections on a regular (non-server) machine.
Regards, NB
EXCUSE, Ash
you're right: i also can not connect with compression enabled ..
Management console reports a connect -- and immediate a disconnect.
But i am maybe on the track why not, HBNETIO seems must be recompiled for that ...
My successful first test with HB_ZLIB_COMPRESSION_SPEED does not change speed for browse in LAN,
at least nothing i can 'feel' ...
regards
Rolf
again Ash,
yep, i can 'see', but not feel, a difference in LAN with compression enabled,
because the traffic is so much below the NIC capability.
If you want to quick test:
you use current branch from GIT ?, then:
line 222 in hbnetio.prg [ in line 221 you see: 'cPassword' ],
replace that upper 'NIL' ( there are two ) with 0 or -1 ...
*BUT* after this mod, you must use a compression value for netio_Connect(),
so a: HB_ZLIB_COMPRESSION: _DEFAULT or _NONE or _SIZE ...
as the default seem: .._DISABLE, which then leads to no connection.
best regards
Rolf
That would allow one RDP user, correct? I need a number of users accessing the server at the same time. Hence my move to NetIO.
Hello Ash,
my tests showed an incredible fast DBSKIP() for LetoDB (*no* filter active) -- no idea why, that's a real master question.
There were in Windows some 'unpredictable' app crashes with LetoDB linked.
A check with valgrind in Linux showed multiple ! memory problems of the Linux app:
180B in 45blocks definitely, 33K!b in 145 blocks possibe lost; invalid read sizes; conditional jump on uninitialised value ...
-- only for just 20 seconds test run. And that was only the client lib, server itself i hadn't checked ...
So actual LetoDB in this state is no option for me.
---
If you don't get HBnetio compression to run for a check with your slow client, ask me ;-)
Ptsarenko fixed the some of the bugs that we reported. If you have precise information about a memory leak, send them to LetoDB forum.
Regards, NB
--
--
You received this message because you are subscribed to the Google
Groups "Harbour Users" group.
Unsubscribe: harbour-user...@googlegroups.com
Web: http://groups.google.com/group/harbour-users
---
You received this message because you are subscribed to the Google Groups "Harbour Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to harbour-user...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.