• Add a dynamic metric like average / max latency to the "Graph" node diagram at the top.
• Make the summary status "tree" 2 level, DB>hosts>services
• Make the summary at the host level have summary figures for the level below (try Max first)
• Make the Graphs filter by parameters for source and target host on the summary row selection
• Make the time bucket for graph query dynamic to plot a sensible number of data points to draw and summarise for the dynamic bucket. How to split your data into a fixed number of buckets (also see date-time functions "{param_name}:grain" function used in the time brush example??)
• Add in average log buffer shipping time to graph as an additional series
• Create a time-range selector that controls the level of aggregation for across the workbook, setting the period for which we average/max across
• Add in "Local log buffer write throughput (MB/s)" and "Log buffer shipping throughput (MB/s)" as averages to the summary view
• try using a spark chart against the summary metrics to give context of bad value vs the norm
• Create an indication that whilst the current status is good, we have had problems in the past (through use of the spark line)
• Add in SERVICE_REPLICATION_STATUS_DETAILS to be viewed via right detail pane

HANA Backup Monitoring MVP

This document will be used to capture the target state for the current iteration.

User story

As a service desk/operations user of the Azure portal without access to SAP tools

OR

SAP BASIS administrator without detailed understanding of the Azure infrastructure

OR

Infrastructure/Cloud administrator without access to the Backup software or SAP tools

I want to visually correlate SAP HANA database backup monitoring data with Azure infrastructure data using Azure Monitor for SAP Solutions to easily troubleshoot issues

The objective is to present the following information through the Azure portal:

Success and failure of data backups with the comment/message from the catalog

Size of backups and duration of backup to complete to support capacity management

This relates to item 3 in the teams discussion

• 1) Fix inconsistency between number of backups on graph and entries in the table for scale-out system
• 2) something seems off with the bar charts - why are some bars overlapping (pic 1) and or look as if there are several, stacked values (pic 2) with different colors/shades?
  See item 4 on this list from Tobias for the pictures
• 3) x axis values often overlapping each other, can they be vertical or only marked on major values?

1. Change the graph visualisation to be instance based rather than node based
2. Simplify the content of the graph-edge to only contain high level state, replication mode in form [Status] in [mode] with [ mode details]
3. Add column name to spark
4. Make the buffer columns current and max separate
5. Try another emoji for non-active status (colour needs to show)
6. Display a message that no HSR for this instance and set a parameter that hides the HSR group
7. Display "Last shipped log time stamp" in details
8. Display "Replay log time stamp" in details
9. Display "Replay backlog size" in details
10. Add in some hint text when the user needs to click on something to activate further details in a hidden object.

Whislt working on collecting tables sizes for the HANA Size project I have identified a constraint with the existing framework.
m_cs_tables view will only return the information required for the top-ten tables when connecting to the tenant. If you connect to the systemdb, you get information relevant to systemdb only. We can resolve this constraint within the existing framework by setting up a provider for each tenant in the system. This has the downside of triggering the collection of duplicate data, since all of the collection queries will be run on all tenants, even though much of the data is host specific and not tenant specific. This could perhaps be resolved by making creating a feature that flagged a check within SapHana.json as “only for systemdb provider” so that only one provider runs that particular check.

• 1) Move some of the detail to a click to view blade to reduce the number of columns in the display to fit within most screens without a scroll (this can be done by defining a link)

Feedback received on workbook using live data:

1. Sync status refers to the the relationship between two nodes to probably should be on the arc line (aka the edge). This can be added to the arc line with an emoji (eg. check mark button and cross mark button)
2. Graph object is best collapsed in a group by default with an option to expand.
3. Make graph object "full" size
4. Graph object should show the state according to the last data collected.
5. Graph object should sit above timeframe selector as it will not be affected by this.
6. Summary tree should have the latest recorded state, however for "seconds behind master" and "async log buffer" we could do a "now (max)" style value or "now (avg)" (also needs to be clear in the columns). Spark graphs can support this.
7. Try colour coding for node
8. Try edge size to be driven by log-ship rate
9. concatenate the Replication mode and operation mode and detailed status into a single column to save space in the summary from mostly empty columns

The HANA metrics used for the following graphs meed to work from the values sampled since the last post, rather than the ever incrementing counters

• query_logbufferwritevsshipping "Log buffer write time vs. shipping time"
• query_logbufferwritethroughputvsshipthroughput "Log buffer write throughput vs. shipping throughput"

Issue

On a scaleout cluster, the backup catalog check will produce duplicated results due to a join on HOSTS

Expected result

There should only be one row per instance, not per host:

sapmon/setup/configureMonitorVM.sh at 6fa91e will hang because it is waiting for user input on apt-get on line 37

# Install PyODBC dependencies
ExecuteCommand "apt-get install g++ unixodbc-dev"

It should be:

# Install PyODBC dependencies
ExecuteCommand "apt-get install -y g++ unixodbc-dev"

Create Line graph for the following 2 metrics:

Local log buffer write throughput (MB/s) = SHIPPED_LOG_BUFFERS_SIZE / SHIPPED_LOG_BUFFERS_DURATION

Log buffer shipping throughput (MB/s) =TOTAL_WRITE_SIZE / TOTAL_WRITE_TIME

Collation of feedback on Backup monitoring workbook

• 1) Decrease the number of decimal places to 0 on [Age of status]
• 2) Change format of [Previous successful backup age] to same as [Age of Status]
• 3) Change [Previous successful backup age] label to [Age of previous successful backup]
• 4) Add a threshold render to the [Number of failed backups] that defaults to the info Icon, and blank if the content starts with a zero
• 5) Summary (Parent) values in the summary grid can just show "fail" if one of the backup type is failed, remove all other values from summary other than total count of failures.
• 6) for "data snapshot", the last one failed - according to "number of failed backups", there must be one that was successful. yet, the "previous successful" column is empty?
• 7) for threshold renderers, i generally prefer to just have the symbol (check or "X") instead of adding "Success" or "Fail" (you can achieve that by leaving the "text" field in the renderer's settings empty)
• 8) Change column label "Last Status" to "Backup status"
• 9) Change column label "Age of Status to "Backup Age"

1. To support data snapshots, we will need to be certain status "Prepared" does not get collected, since the catalog entry gets updated, rather than appended, as we might might miss the final data snapshot status.

We have two base workbooks for review to decide on the base structure for the catalog list -

Flat grid with tiles to filter:
https://portal.azure.com/#@0df5a3ce-fce9-426d-b4d3-1603f3deed39/resource/subscriptions/db5f064b-73c2-4039-8262-f230cc64c659/resourceGroups/rg-rw-experiment-01/providers/microsoft.insights/workbooks/a442c443-8a17-4cde-84ea-6766ce4b808b/overview

Tree grid with top level with a status for each database tenant:
https://portal.azure.com/#@0df5a3ce-fce9-426d-b4d3-1603f3deed39/resource/subscriptions/db5f064b-73c2-4039-8262-f230cc64c659/resourceGroups/sapmon-rg-61164a47e45d81/providers/microsoft.insights/workbooks/9ad23bca-f622-4a61-b93d-502e64f600bd/overview
(not that we have not coded the top level status at this stage, awaiting feedback)

This feedback item relates to No1 in the teams discussion