DataReader vs. DataSet - 未名空间MITBBS历史存档

国际科技财经博客移民网络热点娱乐民生时事公众号

Redian新闻

>未名空间

>DotNet - 窗口里的风景

DataReader vs. DataSet

DataReader vs. DataSet# DotNet - 窗口里的风景

L*r2003-10-21 07:10

1 楼

My words on this:
dataset way is more scalable, especially for busy service(a lot of
connections). But datareader is faster if not that many requests.
hehe, dataset just "looks faster" for busy web serive.
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
adotnetarch031.asp
In all of the preceding tests, we saw that DataReader outperformed DataSet. As
mentioned earlier, the DataReader offers better performance because it avoids
the performance and memory overhead ass

k*i2003-10-21 07:10

2 楼

Personally I like dataset more than datareader.
As you said datareader holds the connection exclusivly.Though the connection
pooling is done by the framework in most of the cases.I still like to share
the connection in my program if possible and it's clearer.

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
As
avoids
connection
it
it

【在 L*******r 的大作中提到】

: My words on this:
: dataset way is more scalable, especially for busy service(a lot of
: connections). But datareader is faster if not that many requests.
: hehe, dataset just "looks faster" for busy web serive.
: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
: adotnetarch031.asp
: In all of the preceding tests, we saw that DataReader outperformed DataSet. As
: mentioned earlier, the DataReader offers better performance because it avoids
: the performance and memory overhead ass

a*y2003-10-21 07:10

3 楼

From my experience, most of the time, DataReader could do the job fine except
sometimes, I need some complex calculation towards a big data set. Probably,
i could achieve the same goal with SQL, but that would make the database too
slow. In this case, dataset is used. We do some sacrifice some memory, but
we saved the database.

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
DataSet.
the
data,
enough
the

【在 k****i 的大作中提到】

: Personally I like dataset more than datareader.
: As you said datareader holds the connection exclusivly.Though the connection
: pooling is done by the framework in most of the cases.I still like to share
: the connection in my program if possible and it's clearer.
:
: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
: As
: avoids
: connection
: it

y*t2003-10-21 07:10

4 楼

I think the key is when you need sort of "real time" data, you have to use
datareader.
and it always compare to dataadaptor instead of dataset.
BTW, is the post really "your words" :)

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
As
avoids
connection
it
it

【在 L*******r 的大作中提到】

L*r2003-10-21 07:10

5 楼

except
Probably,
~~~~~~~~~~~~~~~~~~~?
saved connections ba. :)
connection
share
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
optimized
while
not

【在 a**y 的大作中提到】

: From my experience, most of the time, DataReader could do the job fine except
: sometimes, I need some complex calculation towards a big data set. Probably,
: i could achieve the same goal with SQL, but that would make the database too
: slow. In this case, dataset is used. We do some sacrifice some memory, but
: we saved the database.
:
: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
: DataSet.
: the
: data,

k*i2003-10-21 07:10

6 楼

too
but
I think he was talking about using the in-memory dataset saved huge
server-side oepration at the back-end.
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
the
off
the

【在 L*******r 的大作中提到】

:
: except
: Probably,
: ~~~~~~~~~~~~~~~~~~~?
: saved connections ba. :)
: connection
: share
: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
: optimized
: while

L*r2003-10-21 07:10

7 楼

So the point here is using the "cache" in dataset to do computing, am I right?

【在 k****i 的大作中提到】

:
: too
: but
: I think he was talking about using the in-memory dataset saved huge
: server-side oepration at the back-end.
: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dnbda/html/bd
: the
: off
: the

a*y2003-10-21 07:10

8 楼

I mean, we saved a lot of database operations.

【在 L*******r 的大作中提到】

:
: So the point here is using the "cache" in dataset to do computing, am I right?

L*r2003-10-21 07:10

9 楼

Got you. I remembered some examples. SQL is not good at some sort of
calculation .
Thanks. :)

right?

【在 a**y 的大作中提到】

: I mean, we saved a lot of database operations.

st2003-10-21 07:10

10 楼

nothing is good if it's over used. hehe. I always believe in something in
between. the previous manager likes "cool" M$ stuff, so they wrote sql proc
over 3k lines, which became a disaster to maintain, debug, optimize later.
also M$ sql server replication latency is a know issue, it can hardly be
determined when replication will be done and it causes heavy network traffic

【在 L*******r 的大作中提到】

: Got you. I remembered some examples. SQL is not good at some sort of
: calculation .
: Thanks. :)
:
: right?

k*i2003-10-21 07:10

11 楼

Really?I mean the replication latency problem.I did the same thing in Sybase
replication server b4 and they worked well.Microsoft sql server is similar(if
not exact the same) to sybase.If it has such big latency problem,does it mean
that it's not suitbale for real time system?
btw,who can feed me the relationship(or history) between sybase sql server and
microsoft sql server?I am 2 lazy to do the research lah.

【在 st 的大作中提到】

: nothing is good if it's over used. hehe. I always believe in something in
: between. the previous manager likes "cool" M$ stuff, so they wrote sql proc
: over 3k lines, which became a disaster to maintain, debug, optimize later.
: also M$ sql server replication latency is a know issue, it can hardly be
: determined when replication will be done and it causes heavy network traffic

st2003-10-21 07:10

12 楼

I would suspect so, due to this problem, we use our own middleware for
real time and delayed quote delivery.

【在 k****i 的大作中提到】

: Really?I mean the replication latency problem.I did the same thing in Sybase
: replication server b4 and they worked well.Microsoft sql server is similar(if
: not exact the same) to sybase.If it has such big latency problem,does it mean
: that it's not suitbale for real time system?
: btw,who can feed me the relationship(or history) between sybase sql server and
: microsoft sql server?I am 2 lazy to do the research lah.