alter fragment question
A user on IDS 9.40 (Solaris) had a 64-million-row table round-robin fragmented across two dbspaces, one of which had hit the 16,000,000-page limit. ALTER FRAGMENT ... ADD only accepts one dbspace at a time, and he wanted to avoid two reorgs or an unload/reload needing 65GB+. Replies advised using ALTER FRAGMENT ON TABLE ... INIT FRAGMENT BY ROUND ROBIN (or BY EXPRESSION) listing all four dbspaces at once; the existing dbspaces can be reused since the old partitions are dropped after redistribution, needing only limited extra space. Another poster suggested HPL with FIFOs as a faster unload/reload alternative. The poster didn't report back on testing.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Hi all I have a table that is currently fragmented across two dbspaces and has reached the page limit for a single dbspace (16,000,000 pages). I need to add additional dbspaces to the fragmentation scheme and I was intending to fragment across four spaces. However, it looks like the syntax for alter fragment only allows one dbspace at a time to be added. This is a table that has 64,000,000 rows and is over 700 characters wide. I would rather not reorg it twice to add the two fragments one at a time. Exporting the table, dropping it, recreating it across the 4 fragments, and reloading it is a large effort not to speak of needing over 65gb in filesystem space. I was hoping I was missing something in the documentation but a syntax check of possible statements using Server Studio gives an error for any combination other than a single dbspace. Any thoughts? Thanks Doug MAllister
actv20@gmail.com wrote:
> Hi all
>
> I have a table that is currently fragmented across two dbspaces and has
> reached the page limit for a single dbspace (16,000,000 pages). I need
> to add additional dbspaces to the fragmentation scheme and I was
> intending to fragment across four spaces. However, it looks like the
> syntax for alter fragment only allows one dbspace at a time to be
> added.
> This is a table that has 64,000,000 rows and is over 700 characters
> wide. I would rather not reorg it twice to add the two fragments one at
> a time.
> Exporting the table, dropping it, recreating it across the 4 fragments,
> and reloading it is a large effort not to speak of needing over 65gb in
> filesystem space.
> I was hoping I was missing something in the documentation but a syntax
> check of possible statements using Server Studio gives an error for any
> combination other than a single dbspace.
>
> Any thoughts?
>
> Thanks
>
> Doug MAllister
>
Hi,
Version of IDS?
Platform?
Pretty large table there :P
What is your existing schema (dbschema -d dbname -t tabname -ss)?
It would be useful to know if it is round robin or actually by expression.
Are the 4 "new fragments" unused at present or are 2 of the fragments
"the existing" ones?
The following snippet from the V10 Information center (centre surely)
suggests some possibilities :
CREATE TABLE account (col1 INT, col2 INT)
FRAGMENT BY ROUND ROBIN IN dbsp1, dbsp2;
ALTER FRAGMENT ON TABLE account
INIT FRAGMENT BY EXPRESSION
col1 < 0 IN dbsp1,
col2 >= 0 IN dbsp2;
with the additional (expected) caveat ...
"If an existing dbspace is full when you redefine a fragmentation
strategy, you must not use it in the new fragmentation strategy.!"
I would suggest you do experiments outside of Server Studio with simple,
small tests using dbaccess and .. a small, fragmented table :)
In fact ... the following all appear to work for me ...
create table tab_two_frag (col1 int, col2 char(25) )
fragment by round robin in dbspace_f1, dbspace_f2;
alter fragment on table tab_two_frag
init fragment by round robin in dbspace_f1, dbspace_f2, dbspace_f3,dbspace_f4;
drop table tab_two_frag;
create table tab_two_frag (col1 int, col2 char(25) )
fragment by round robin in dbspace_f1, dbspace_f2;
alter fragment on table tab_two_frag
init fragment by round robin in dbspace_f3, dbspace_f4, dbspace_f5,dbspace_f6;
drop table tab_two_frag;
create table tab_two_frag (col1 int, col2 char(25) )
fragment by expression
col1 < 0 in dbspace_f1,col1 >=0 in dbspace_f2;
alter fragment on table tab_two_fraginit fragment by expression
col1 < 0 in dbspace_f3,
col1 < 1000000 and col1 >= 0 in dbspace_f4,
col1 < 2000000 and col1 >= 1000000 in dbspace_f5,
col1 >= 2000000 in dbspace_f6;
So .. there ya go
actv20@gmail.com wrote:
> Hi all
>
> I have a table that is currently fragmented across two dbspaces and has
> reached the page limit for a single dbspace (16,000,000 pages). I need
> to add additional dbspaces to the fragmentation scheme and I was
> intending to fragment across four spaces. However, it looks like the
> syntax for alter fragment only allows one dbspace at a time to be
> added.
Don't ADD a fragment, reinit the entire table's fragmentation scheme with:
ALTER FRAGMENT ON TABLE mytableINIT FRAGMENT BY EXPRESSION
year <= 2000 IN dbspace1,
year = 2001 IN dbspace2,
year = 2002 IN dbspace2,
....
REMAINDER in dbspaceN;
Art S. Kagel
> This is a table that has 64,000,000 rows and is over 700 characters
> wide. I would rather not reorg it twice to add the two fragments one at
> a time.
> Exporting the table, dropping it, recreating it across the 4 fragments,
> and reloading it is a large effort not to speak of needing over 65gb in
> filesystem space.
> I was hoping I was missing something in the documentation but a syntax
> check of possible statements using Server Studio gives an error for any
> combination other than a single dbspace.
>
> Any thoughts?
>
> Thanks
>
> Doug MAllister
>
Sorry, I wasn't too clear the first time around.
9.40.fc1 is the version on Solaris.
I understand about init fragment caveats, in this case alter fragment
has syntax to ADD a dbspace without doing an init, and yes it is round
robin.
You do not want to see the whole schema, over 1000 tables, but here is
the specific table sql (not mine, blame third-party commercial
application)
create table 'gim2usr'.ltxns (
ltxn_code SERIAL not null,
lt_refer INT,
cl_id CHAR(12),
lot_code INT,
ho_code INT,
txn_date DATE,
set_date DATE,
org_set_date DATE,
sys_date DATE,
symbol CHAR(12),
brok_id CHAR(8),
cust_id CHAR(8),
txn_code CHAR(2),
txn_type CHAR(1),
tra_type CHAR(1),
account CHAR(1),
bint_account CHAR(1),
bint_accr_day INT,
shr_face DECIMAL(18,4),
u_price FLOAT,
pcost MONEY(16,2),
pdate DATE,
amt_mkt MONEY(16,2),
commis MONEY(16,2),
bond_int MONEY(16,2),
sec_fee MONEY(16,2),
fx_tax MONEY(16,2),
oth_fee MONEY(16,2),
fx_fee MONEY(16,2),
org_face DECIMAL(18,4),
org_cost MONEY(16,2),
parent INT,
mkt_rate FLOAT,
dstatus CHAR(1),
can_date DATE,
base_rate FLOAT,
trd_cur CHAR(12),
stl_cur CHAR(12),
client_cur CHAR(12),
m_exc_rate FLOAT,
firm_amt_cycost MONEY(16,2),
client_amt_cycost MONEY(16,2),
stl_amt MONEY(16,2),
stl_bint MONEY(16,2),
firm_bint_cycost MONEY(16,2),
client_bint_cycost MONEY(16,2),
firm_amt MONEY(16,2),
client_amt MONEY(16,2),
firm_pcost MONEY(16,2),
client_pcost MONEY(16,2),
cst_conf_stat CHAR(1),
grp_date DATE,
grp_refer INT,
org_date DATE,
org_refer INT,
owner CHAR(8),
notice_1 CHAR(1),
notice_2 CHAR(1),
broker_inst CHAR(6),
status CHAR(1),
ext_ref CHAR(1),
neg_accrue CHAR(1),
cstatus CHAR(1),
perfcd CHAR(8),
perf_posted DATE,
rcv_date DATE,
brok_reason CHAR(10),
dcontrol CHAR(1),
cmnt CHAR(6),
comm_type CHAR(1),
st_txn_refer CHAR(9),
int_at_mat MONEY(16,2),
fwd_type CHAR(12),
index_ratio FLOAT,
ws_code CHAR(2),
ext_lot CHAR(15),
registration_code CHAR(8),
m_expo_rate FLOAT,
expo_amt MONEY(16,2),
expo_pcost MONEY(16,2),
expo_cur CHAR(12),
trd_blk_ref CHAR(20),
m_client_rate FLOAT,
d_client_rate FLOAT,
m_firm_rate FLOAT,
d_firm_rate FLOAT,
d_expo_rate FLOAT,
d_exc_rate FLOAT,
blm_mstr_tkt INT,
ext_grp INT,
fwd_id CHAR(12),
can_user CHAR(8),
broker_buy_inst CHAR(6),
ex_date DATE,
fr_amt_cycost_set MONEY(16,2),
cl_amt_cycost_set MONEY(16,2),
fr_bint_cycost_set MONEY(16,2),
cl_bint_cycost_set MONEY(16,2),
fr_pcost_set MONEY(16,2),
cl_pcost_set MONEY(16,2),
cstatus_set CHAR(1),
bond_equ_yld FLOAT,
can_refer INT,
alloc_seq_no INT,
dtc_ii_status CHAR(1),
pbond_int MONEY(16,2)
)
fragment by round robin in datadbs10,datadbs11
extent size 2000000 next size 600000
lock mode row;
create index 'gim2usr'.dtc_match on 'gim2usr'.ltxns
(
txn_date,
txn_code,
symbol,
cl_id
)
in datadbs6;
create index 'gim2usr'.lt_idx1 on 'gim2usr'.ltxns
(
cl_id
)
in datadbs5;
create index 'gim2usr'.lt_idx2 on 'gim2usr'.ltxns
(
symbol
)
in datadbs6;
create unique index 'gim2usr'.lt_refer on 'gim2usr'.ltxns
(
sys_date,
lt_refer
)
in datadbs4;
create index 'gim2usr'.rds01 on 'gim2usr'.ltxns
(
ho_code,
lot_code,
rcv_date,
txn_date
)
in datadbs7;
These are not 4 NEW fragments, I only propose to ADD 2 new fragments to
the scheme. I can add a single fragment at a time but would rather not
do 2 reorgs on 64 million rows
b-t-w, Server Studio is useful as a syntax checker for sql, not
necessarily to actually run this kind of sql.
TBP wrote:
> actv20@gmail.com wrote:
> > Hi all
> >
> > I have a table that is currently fragmented across two dbspaces and has
> > reached the page limit for a single dbspace (16,000,000 pages). I need
> > to add additional dbspaces to the fragmentation scheme and I was
> > intending to fragment across four spaces. However, it looks like the
> > syntax for alter fragment only allows one dbspace at a time to be
> > added.
> > This is a table that has 64,000,000 rows and is over 700 characters
> > wide. I would rather not reorg it twice to add the two fragments one at
> > a time.
> > Exporting the table, dropping it, recreating it across the 4 fragments,
> > and reloading it is a large effort not to speak of needing over 65gb in
> > filesystem space.
> > I was hoping I was missing something in the documentation but a syntax
> > check of possible statements using Server Studio gives an error for any
> > combination other than a single dbspace.
> >
> > Any thoughts?
> >
> > Thanks
> >
> > Doug MAllister
> >
> Hi,
>
> Version of IDS?
> Platform?
>
> Pretty large table there :P
>
> What is your existing schema (dbschema -d dbname -t tabname -ss)?
> It would be useful to know if it is round robin or actually by expression.
>
> Are the 4 "new fragments" unused at present or are 2 of the fragments
> "the existing" ones?
>
> The following snippet from the V10 Information center (centre surely)
> suggests some possibilities :
>
> CREATE TABLE account (col1 INT, col2 INT)
> FRAGMENT BY ROUND ROBIN IN dbsp1, dbsp2;
> ALTER FRAGMENT ON TABLE account
> INIT FRAGMENT BY EXPRESSION
> col1 < 0 IN dbsp1,
> col2 >= 0 IN dbsp2;>
> with the additional (expected) caveat ...
>
> "If an existing dbspace is full when you redefine a fragmentation
> strategy, you must not use it in the new fragmentation strategy.!"
>
> I would suggest you do experiments outside of Server Studio with simple,
> small tests using dbaccess and .. a small, fragmented table :)
>
> In fact ... the following all appear to work for me ...
>
> create table tab_two_frag (col1 int, col2 char(25) )
> fragment by round robin in dbspace_f1, dbspace_f2;
> alter fragment on table tab_two_frag
> init fragment by round robin in dbspace_f1, dbspace_f2, dbspace_f3,> dbspace_f4;
>
> drop table tab_two_frag;>
> create table tab_two_frag (col1 int, col2 char(25) )
> fragment by round robin in dbspace_f1, dbspace_f2;
> alter fragment on table tab_two_frag
> init fragment by round robin in dbspace_f3, dbspace_f4, dbspace_f5,> dbspace_f6;
>
> drop table tab_two_frag;>
> create table tab_two_frag (col1 int, col2 char(25) )
> fragment by expression
> col1 < 0 in dbspace_f1,> col1 >=0 in dbspace_f2;
> alter fragment on table tab_two_frag> init fragment by expression
> col1 < 0 in dbspace_f3,
> col1 < 1000000 and col1 >= 0
My assumption is that ADD DBSPACE syntax will allow me to keep what I
already have and add an additional dbspace. Refragmenting forces me
into entirely new dbspaces. While that may be a good idea, with 64gb
already allocated, I would need to find another 128gb and then end up
with the original 64gb not used. (Finding 128gb on a SAN is not like
walking down to Circuit City and plunking down $99 for a 300gb drive).
I think my option here is to add a single dbspace, bringing my total to
96gb and 3 fragments then wait a year for the table to hit the limit
again and do another single dbspace add.
Doug...
Art S. Kagel wrote:
> actv20@gmail.com wrote:
> > Hi all
> >
> > I have a table that is currently fragmented across two dbspaces and has
> > reached the page limit for a single dbspace (16,000,000 pages). I need
> > to add additional dbspaces to the fragmentation scheme and I was
> > intending to fragment across four spaces. However, it looks like the
> > syntax for alter fragment only allows one dbspace at a time to be
> > added.
>
> Don't ADD a fragment, reinit the entire table's fragmentation scheme with:
>
> ALTER FRAGMENT ON TABLE mytable> INIT FRAGMENT BY EXPRESSION
> year <= 2000 IN dbspace1,
> year = 2001 IN dbspace2,
> year = 2002 IN dbspace2,
> ....
> REMAINDER in dbspaceN;
>
> Art S. Kagel
>
> > This is a table that has 64,000,000 rows and is over 700 characters
> > wide. I would rather not reorg it twice to add the two fragments one at
> > a time.
> > Exporting the table, dropping it, recreating it across the 4 fragments,
> > and reloading it is a large effort not to speak of needing over 65gb in
> > filesystem space.
> > I was hoping I was missing something in the documentation but a syntax
> > check of possible statements using Server Studio gives an error for any
> > combination other than a single dbspace.
> >
> > Any thoughts?
> >
> > Thanks
> >
> > Doug MAllister
> >
actv20@gmail.com wrote:
> Sorry, I wasn't too clear the first time around.
>
> 9.40.fc1 is the version on Solaris.
> I understand about init fragment caveats, in this case alter fragment
> has syntax to ADD a dbspace without doing an init, and yes it is round
> robin.
> You do not want to see the whole schema, over 1000 tables, but here is
> the specific table sql (not mine, blame third-party commercial
> application)
>
> create table 'gim2usr'.ltxns (
> ltxn_code SERIAL not null,
<snip really big schema>
> pbond_int MONEY(16,2)
> )
> fragment by round robin in datadbs10,datadbs11
> extent size 2000000 next size 600000
> lock mode row;
>
<snip some more schemas>
>
> These are not 4 NEW fragments, I only propose to ADD 2 new fragments to
> the scheme. I can add a single fragment at a time but would rather not
> do 2 reorgs on 64 million rows
>
Tell more about the "reorgs".
Have a look at the following :
create table tab_two_frag (col1 int, col2 char(25) )
fragment by round robin in dbspace_f1, dbspace_f2;
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
insert into tab_two_frag values (1,"asdasdasd");
alter fragment on table tab_two_fragadd dbspace_f3;
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
insert into tab_two_frag values (2,"qweqweqweqwe");
alter fragment on table tab_two_fragadd dbspace_f4;
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
insert into tab_two_frag values (3,"zxczxczxczxc");
> oncheck -pt doug_a:tab_two_frag | egrep -i "Dbspace|rows" Table fragment partition dbspace_f1 in DBspace dbspace_f1
Number of rows 13
Table fragment partition dbspace_f2 in DBspace dbspace_f2
Number of rows 9
Table fragment partition dbspace_f3 in DBspace dbspace_f3
Number of rows 7
Table fragment partition dbspace_f4 in DBspace dbspace_f4
Number of rows 4
> b-t-w, Server Studio is useful as a syntax checker for sql, not
> necessarily to actually run this kind of sql.
>
Carefull here... test it first on a test box you may get an unexpected rebuild of the table what was the original fragmentation schema?? you may have to modify that first??? or how are you preventing rows being inserted into the 'full' fragement. reloading.... you may look at the high perf loader to reload the table; AKA nuke obstacle tool. use fifos so you do not have to allocate a whole lot of filesystem space; HPL in express mode that is(rowsize is smaller then a page...) so create new table with fragmentation strategy you want, unload data using hpl to a device array containing fifos load data into new table using hpl reading from dev array containing fifos; use about # cpus of fifos. know for a fact that this will be faster then alter fragment...; you can get alter fragment... to execute fast by tweaking the engine a lot, you can get close to the speed of HPL but you will not beat it... NOTE check your data in case the os mucks up. Superboer.
actv20@gmail.com wrote:
> My assumption is that ADD DBSPACE syntax will allow me to keep what I
> already have and add an additional dbspace. Refragmenting forces me
> into entirely new dbspaces. While that may be a good idea, with 64gb
NO! The ALTER FRAGMENT ... INIT can name the two existing fragments and two
additional ones. All that's going to happen is that the existing rows will
be redistributed to the four new partitions then the original partitions
will be dropped. The engine knows it's going to be able to drop the
original partitions so it WILL allow you to use the same dbspaces, even in
versions prior to 10.00 which allows multiple fragments in a dbspace. You
will need space in the original two dbspaces for about 1/4 of the data in
both existing dbspaces taken together. So, based on your other post, you
should be able to:
ALTER FRAGMENT ON TABLE mytableINIT FRAGMENT BY ROUND ROBIN IN datadbs10, datadbs11, datadbs12, datadbs13;
Doug, I'm usually the last to berate posters, but you're approaching
DenseVille here. You asked for a solution, we gave you one, instead of
telling us that it won't work just try it!
Art S. Kagel
> already allocated, I would need to find another 128gb and then end up
> with the original 64gb not used. (Finding 128gb on a SAN is not like
> walking down to Circuit City and plunking down $99 for a 300gb drive).
> I think my option here is to add a single dbspace, bringing my total to
> 96gb and 3 fragments then wait a year for the table to hit the limit
> again and do another single dbspace add.
> Doug...
>
> Art S. Kagel wrote:
>
>>actv20@gmail.com wrote:
>>
>>>Hi all
>>>
>>>I have a table that is currently fragmented across two dbspaces and has
>>>reached the page limit for a single dbspace (16,000,000 pages). I need
>>>to add additional dbspaces to the fragmentation scheme and I was
>>>intending to fragment across four spaces. However, it looks like the
>>>syntax for alter fragment only allows one dbspace at a time to be
>>>added.
>>
>>Don't ADD a fragment, reinit the entire table's fragmentation scheme with:
>>
>>ALTER FRAGMENT ON TABLE mytable>>INIT FRAGMENT BY EXPRESSION
>> year <= 2000 IN dbspace1,
>> year = 2001 IN dbspace2,
>> year = 2002 IN dbspace2,
>>....
>> REMAINDER in dbspaceN;
>>
>>Art S. Kagel
>>
>>
>>>This is a table that has 64,000,000 rows and is over 700 characters
>>>wide. I would rather not reorg it twice to add the two fragments one at
>>>a time.
>>>Exporting the table, dropping it, recreating it across the 4 fragments,
>>>and reloading it is a large effort not to speak of needing over 65gb in
>>>filesystem space.
>>>I was hoping I was missing something in the documentation but a syntax
>>>check of possible statements using Server Studio gives an error for any
>>>combination other than a single dbspace.
>>>
>>>Any thoughts?
>>>
>>>Thanks
>>>
>>>Doug MAllister
>>>
>
>
Well, thanks for the help here, I'm just glad I didn't make a comment
about Stuart Litel, I would have really been flamed.
Art S. Kagel wrote:
> actv20@gmail.com wrote:
> > My assumption is that ADD DBSPACE syntax will allow me to keep what I
> > already have and add an additional dbspace. Refragmenting forces me
> > into entirely new dbspaces. While that may be a good idea, with 64gb
>
> NO! The ALTER FRAGMENT ... INIT can name the two existing fragments and two
> additional ones. All that's going to happen is that the existing rows will
> be redistributed to the four new partitions then the original partitions
> will be dropped. The engine knows it's going to be able to drop the
> original partitions so it WILL allow you to use the same dbspaces, even in
> versions prior to 10.00 which allows multiple fragments in a dbspace. You
> will need space in the original two dbspaces for about 1/4 of the data in
> both existing dbspaces taken together. So, based on your other post, you
> should be able to:
>
> ALTER FRAGMENT ON TABLE mytable> INIT FRAGMENT BY ROUND ROBIN IN datadbs10, datadbs11, datadbs12, datadbs13;
>
> Doug, I'm usually the last to berate posters, but you're approaching
> DenseVille here. You asked for a solution, we gave you one, instead of
> telling us that it won't work just try it!
>
> Art S. Kagel
>
> > already allocated, I would need to find another 128gb and then end up
> > with the original 64gb not used. (Finding 128gb on a SAN is not like
> > walking down to Circuit City and plunking down $99 for a 300gb drive).
> > I think my option here is to add a single dbspace, bringing my total to
> > 96gb and 3 fragments then wait a year for the table to hit the limit
> > again and do another single dbspace add.
> > Doug...
> >
> > Art S. Kagel wrote:
> >
> >>actv20@gmail.com wrote:
> >>
> >>>Hi all
> >>>
> >>>I have a table that is currently fragmented across two dbspaces and has
> >>>reached the page limit for a single dbspace (16,000,000 pages). I need
> >>>to add additional dbspaces to the fragmentation scheme and I was
> >>>intending to fragment across four spaces. However, it looks like the
> >>>syntax for alter fragment only allows one dbspace at a time to be
> >>>added.
> >>
> >>Don't ADD a fragment, reinit the entire table's fragmentation scheme with:
> >>
> >>ALTER FRAGMENT ON TABLE mytable> >>INIT FRAGMENT BY EXPRESSION
> >> year <= 2000 IN dbspace1,
> >> year = 2001 IN dbspace2,
> >> year = 2002 IN dbspace2,
> >>....
> >> REMAINDER in dbspaceN;
> >>
> >>Art S. Kagel
> >>
> >>
> >>>This is a table that has 64,000,000 rows and is over 700 characters
> >>>wide. I would rather not reorg it twice to add the two fragments one at
> >>>a time.
> >>>Exporting the table, dropping it, recreating it across the 4 fragments,
> >>>and reloading it is a large effort not to speak of needing over 65gb in
> >>>filesystem space.
> >>>I was hoping I was missing something in the documentation but a syntax
> >>>check of possible statements using Server Studio gives an error for any
> >>>combination other than a single dbspace.
> >>>
> >>>Any thoughts?
> >>>
> >>>Thanks
> >>>
> >>>Doug MAllister
> >>>
> >
> >