How to use least squares methods for resource assessment and monitoring of Oracle databases

In daily activities, Oracle database administrators need to track a significant number of metrics in terms of bandwidth, load, growth of DBMS, DBMS hosts, disk subsystems, etc. All this data is carefully collected and stored by monitoring systems. So that the information does not lie dead weight, it can be used for forecasting in terms of sizing issues and even proactive monitoring. In this post we will demonstrate how to do it.



Take the Oracle Cloud Control. In the SYSMAN schema in the MGMT_METRICS_1DAY , we store the value of the metric “Filesystem Space Available (MB)” for a certain host with TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C' .

 SELECT TO_CHAR (CAST (ROLLUP_TIMESTAMP AS DATE), 'YYYY-MM-DD"T"HH24:MI:SS'),        ROUND (VALUE_AVERAGE, 2) AVG_FREE_SPACE_INM   FROM SYSMAN.MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; 

The graph below shows that for three months, from August 14, 2018 to November 15, 2018, free space in the / u01 file system decreased by 1.5TB - from 4.1 TB to 2.6 TB
')


To determine the time for which we have enough of the remaining 2.6 TB, we turn to the functions of linear regression.

In Oracle, the REGR_SLOPE(x,y) function is used to calculate the slope of the regression line. The slope function of the regression line REGR_SLOPE is determined by the ratio of the covariance of the sets x and y to the variance of the set y:

 REGR_SLOPE(x,y) = COVAR_POP(x,y) / VAR_POP(y) 

To calculate the interception of the Y axis, the function REGR_INTERCEPT(x,y) . The intercept function of the Y axis REGR_INTERCEPT is determined by the difference in the mean value of the set x and the product of the slope of the regression line and the mean value of the set y:

 REGR_INTERCEPT(x,y) = AVG(x) – REGR_SLOPE(x, y) * AVG(y) 

To calculate the R-square or coefficient of determination, the function REGR_R2 (x, y) is used, which is not defined at zero variance y, is equal to one at zero variance x and non-zero dispersion y. And in the case of a positive variance of x and a non-zero variance, y is equal to the square of the correlation of x and y:

 NULL if VAR_POP(y)  = 0 1 if VAR_POP(x)  = 0 and VAR_POP(y) != 0 POWER(CORR(expr1,expr),2) if VAR_POP(x)  > 0 and VAR_POP(y)  != 0 

We would describe the coefficient of determination as a value characterizing the percentage of the described variations by the model. If the coefficient of determination is 1, then our model describes 100% of variations and our forecast will be as accurate as possible.

We suggest to take as x the time difference in which the metric was obtained and the current time - ROLLUP_TIMESTAMP-SYSDATE. As y, we take the average free disk space / u01 in MB - “Filesystem Space Available (MB)”. In this case, using the REGR_INTERCEPT function (ROLLUP_TIMESTAMP-SYSDATE, VALUE_AVERAGE) as a group, we can estimate how many days the / u01 disk space runs out.

In expanded form, the value of the function of the slope of the regression line, the interception of the axis and the coefficient of determination is obtained by the query:

 SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; (VALUE_AVERAGE, SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; ROUND (VALUE_AVERAGE, SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; (VALUE_AVERAGE, SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; ROUND (VALUE_AVERAGE SELECT   COVAR_POP (ROLLUP_TIMESTAMP - SYSDATE, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) / VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) REGR_SLOPE,        AVG (ROLLUP_TIMESTAMP)- (COVAR_POP (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE, 2), 0)))*AVG (NULLIF (ROUND(VALUE_AVERAGE, 2), 0))           REGR_INTERCEPT,        AVG (SYSDATE - ROLLUP_TIMESTAMP)-(COVAR_POP(SYSDATE - ROLLUP_TIMESTAMP,NULLIF(ROUND(VALUE_AVERAGE,2),0))/VAR_POP(NULLIF(ROUND(VALUE_AVERAGE,2),0)))*AVG(NULLIF(ROUND(VALUE_AVERAGE,2),0)))      REGR_INTERCEPT_ABS,        CASE           WHEN VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) = 0           THEN              NULL           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) = 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              1           WHEN     VAR_POP (SYSDATE - ROLLUP_TIMESTAMP) > 0                AND VAR_POP (NULLIF (ROUND (VALUE_AVERAGE, 2), 0)) != 0           THEN              POWER (CORR (SYSDATE - ROLLUP_TIMESTAMP, NULLIF (ROUND (VALUE_AVERAGE, 2), 0)),2)        END           REGR_R2   FROM MGMT_METRICS_1DAY  WHERE TARGET_GUID = '6B1E3AFA92B3EA29AD73BB87432C084C'        AND METRIC_GUID = HEXTORAW ('E8838C71E687BF0A9E02FFACC0C9AC80')        AND KEY_VALUE = '/u10' ORDER BY ROLLUP_TIMESTAMP; 

Or using the REGR_SLOPE , REGR_INTERCEPT and REGR_R2 :

 SELECT REGR_SLOPE(SYSDATE-ROLLUP_TIMESTAMP, NULLIF(ROUND(VALUE_AVERAGE,2),0)) RSLP,      REGR_INTERCEPT(SYSDATE-ROLLUP_TIMESTAMP, NULLIF(ROUND(VALUE_AVERAGE,2),0)) RINSP,       REGR_R2(SYSDATE-ROLLUP_TIMESTAMP, NULLIF(ROUND(VALUE_AVERAGE,2),0)) RSQR      FROM MGMT_METRICS_1DAY      WHERE TARGET_GUID='6B1E3AFA92B3EA29AD73BB87432C084C'      AND METRIC_GUID=HEXTORAW('E8838C71E687BF0A9E02FFACC0C9AC80') AND KEY_VALUE='/u10'      ORDER BY ROLLUP_TIMESTAMP; 

 REGR_SLOPE = -0.00005 REGR_INTERCEPT = 149.46 REGR_R2 = 0.97 

In our case, the coefficient of determination is close to 1, and we see that after about 149 days on the / u10 file system will end.

We can use this method to estimate the time for which we have enough disk space for hosts of productive bases. In our case, the hosts of the production databases are included in the group with METRIC_GUID='E8838C71E687BF0A9E02FFACC0C9AC80' .

Create a User Definded Type T_TYPE with the fields we need for the output: host name, file system name, regression line slope, the number of days after which the space on the file system will end and the coefficient of determination.

 CREATE OR REPLACE TYPE T_TYPE AS OBJECT(TARGET_NAME VARCHAR2(256), KEY_VALUE VARCHAR2(256), RSLP NUMBER,RINSP NUMBER,RSQR NUMBER); 

Create a Nested Table Type R_TYPE based on T_TYPE:

 CREATE OR REPLACE TYPE R_TYPE AS TABLE OF T_TYPE; 

We create a package that includes the GET_VALUES function to extract data from MGMT_METRICS_1DAY and the procedure for sending the received data by mail SEND_VALUES .

 CREATE OR REPLACE PACKAGE EST_FS_EXHAUST IS 

For the GET_VALUES function GET_VALUES input parameters will be the V_GN variable with the name of the host group and the value of the coefficient of determination V_RSQ .

 FUNCTION GET_VALUES(V_GN VARCHAR2, V_RSQ NUMBER) RETURN R_TYPE; 

For the SEND_VALUES procedure SEND_VALUES input parameters will be the V_GN and V_RSQ variables, similar to the GET_VALUES function GET_VALUES , plus an email, to which we plan to send our mini-report.

 PROCEDURE SEND_VALUES(V_GN VARCHAR2, V_RSQ NUMBER, V_MAIL VARCHAR2); END EST_FS_EXHAUST; 

In the package body we define the function GET_VALUES and the procedure SEND_VALUES

 CREATE OR REPLACE PACKAGE BODY EST_FS_EXHAUST IS FUNCTION GET_VALUES(V_GN VARCHAR2, V_RSQ NUMBER) RETURN R_TYPE AS  V_REC R_TYPE; BEGIN SELECT T_TYPE( M.TARGET_NAME, D.KEY_VALUE, ROUND(REGR_SLOPE(D.ROLLUP_TIMESTAMP-SYSDATE, NULLIF(ROUND(D.VALUE_AVERAGE,2),0)),0), ROUND((ABS(REGR_INTERCEPT(D.ROLLUP_TIMESTAMP-SYSDATE, NULLIF(ROUND(D.VALUE_AVERAGE,2),0)))),0), ROUND(REGR_R2(D.ROLLUP_TIMESTAMP-SYSDATE, NULLIF(ROUND(D.VALUE_AVERAGE,2),0)),0)) BULK COLLECT INTO V_REC FROM MGMT_METRICS_1DAY D, MGMT_TARGETS M, MGMT_TARGET_MEMBERSHIPS G WHERE M.TARGET_GUID=G.MEMBER_TARGET_GUID AND M.TARGET_GUID=D.TARGET_GUID AND G.COMPOSITE_TARGET_NAME=V_GN AND METRIC_GUID=HEXTORAW('E8838C71E687BF0A9E02FFACC0C9AC80') GROUP BY M.TARGET_NAME, D.KEY_VALUE HAVING REGR_R2(D.ROLLUP_TIMESTAMP-SYSDATE, NULLIF(ROUND(D.VALUE_AVERAGE,2),0)) > V_RSQ      ; RETURN V_REC; END GET_VALUES; PROCEDURE SEND_VALUES(V_GN VARCHAR2, V_RSQ NUMBER, V_MAIL VARCHAR2) IS V_REC R_TYPE; MSG VARCHAR2(2048):=''; BEGIN V_REC:= GET_VALUES(V_GN,V_RSQ); FOR I IN V_REC.FIRST..V_REC.LAST LOOP MSG:=CHR(10)||MSG||' Host '||V_REC(I).TARGET_NAME||' filesystem '||V_REC(I).KEY_VALUE||' will be exhausted in '||V_REC(I).RINSP||' days'|| CHR(9)||CHR(10); END LOOP; EXECUTE IMMEDIATE 'ALTER SESSION SET smtp_out_server = ''mail_server'''; UTL_MAIL.SEND(SENDER => 'monitoring@yourmail.com',       RECIPIENTS => V_MAIL,          SUBJECT => 'Test Mail',          MESSAGE => MSG,        MIME_TYPE => 'text; charset=us-ascii'); END; END EST_FS_EXHAUST; / 

Suppose we want to get the time after which the place in the file systems of the group of productive hosts 'prod_hosts' ends up with a determination coefficient greater than 0.5:

 begin EST_FS_EXHAUST.SEND_VALUES('prod_hosts',0.5,'operator@yourdomain.com'); end; / PL/SQL procedure successfully completed. 

As a result, the message arrives in the mail:

Host host1 filesystem / u51 will be exhausted in 342 days
Host host2 filesystem / u40 will be exhausted in 236 days
Host host3 filesystem / u20 / redo01 will be exhausted in 1100310 days
Host host4 filesystem / u10 will be exhausted in 150 days
Host host4 filesystem / u01 / integration will be exhausted in 75080 days
Host host4 filesystem / u01 / app will be exhausted in 135 days
Host host5 filesystem / u30 / redo01 will be exhausted in 62252596 days
Host host6 filesystem / u01 will be exhausted in 260 days
Host host7 filesystem / u99 will be exhausted in 1038 days

Pay attention to the file systems / u20 / redo01 and / u30 / redo01 - here are the REDO LOGS and space is not spent. Our model predicted that the place in / u20 / redo01 on host3 will end in 2990 years, and in / u30 / redo01 on host5 - in 169164 years. In both cases, the regression line slope is less than –1.

The use of this method can be extended - for example, estimate the time until the place is exhausted in FAST RECOVERY AREA, in the tablespaces of the database and other areas.