Skip to Content

A recent SAP HANA Cloud downtime announcement on SCN caught my eye.

Dear SAP HANA Cloud customers,

Please be reminded that there will be a planned downtime for https://hana.ondemand.com on April 11th, between 05:00 – 06:00 UTC due to an infrastructure update.


Impact

Development operations like deployment, starting and stopping of applications will not be possible.

The following platform services will be down during the update:

    • Document Service
    • HANA DB

The other underlying platform services might experience short outages.

Running applications will be accessible.

Note: I closely follow the SAP HANA Cloud downtime announcements on SCN and the resulting updated release notes. I always look forward to seeing how the platform has evolved and I’m impressed that the team is keeping up with its two-week release cycle.  The public communication of such events is also an important sign of transparency vis-a-vis the ecosystem. 

The associated deployment was completed successfully but the initial announcement still irritated me.  Usually, such announcements contain statements like this one –  “The other underlying platform services might experience short outages.“ but this time two critical services were down during the update. I have no idea how long the downtime really lasted but it was announced as lasting one hour.   Running applications might be accessible but what if they used the document or HANA DB service? Would such applications still work? I don’t know whether the downtime announcement was just unclear or whether applications using the affected services were also impacted.  I also have no idea have many applications are currently productive on the platform or how many of these applications use these two services.

So many unanswered questions were disturbing. I was recently doing research on cloud SLAs and I started wondering about the downtime announcement and SAP’s SLAs for the offering.  Could the downtime of these two services be considered a SLA violation?

I looked for the SLAs of the HANA Cloud and only found an old SAP NetWeaver Cloud Supplemental Terms and Conditions from 10/2012 which contains this clause:

5. Platform Availability

With respect to the Platform (includes compute, structured storage, unstructured storage, bandwidth, connectivity connector in the Cloud), SAP warrants at least ninety-nine point nine percent (99.9%) System Availability over any calendar month. Should SAP fail to achieve ninety-nine point nine percent (99.9%) System Availability over a calendar month, Customer shall have the right to receive a credit equal to two percent (2%) of its subscription fees for the Service (for Platform fees only) for that month, for each one percent (1%) (or portion thereof) by which SAP fails to achieve such level, up to one hundred percent (100%) of the fees for such month.

Note: This paragraph really isn’t detailed enough for a cloud service. The description provided by the HP Cloud group regarding their SLAs is much easier to understand.

So, we have an availability of 99,9%. What is the acceptable downtime in a month to maintain the SLA? I used the Ez (Easy) Uptime Calculator to calculate it.

/wp-content/uploads/2013/04/image001_206905.jpg

Just a downtime of 43.2 minutes would be permitted in a month. If the HANA Cloud services were really down for an hour, then it appears that we might have a SLA violation – especially since the “Platform” described in the availability clause in the ToC is defined as including “structured storage, unstructured storage”.

Conclusion

Why am I even bringing up the topic? The deployment was successful and everything is fine. 

There is currently a heated debate about the quality of the support provided for SAP’s cloud applications.  In the comment stream to this blog,  Sina Moatamed makes a very important point:  “First challenge is that SAP is no longer in the software business.  Cloud is about Service Delivery.  The entire engagement has to be from that lens.” Although the blog in question focuses on metrics associated with support for SAP’s cloud applications, Sina’s point is very important and represents a shift for SAP.

Maybe I’m making a mountain out of a molehill and the two services weren’t even down for an hour. Yet as SAP moves more rapidly into the cloud space, it must be very aware that the market is closely watching its actions. SAP’s image and a judgement on its degree of professionalism regarding its cloud offering are still in flux – especially in comparison to other vendors / competitors.  Despite the fact that SAP’s cloud / platform strategy has many moving parts, there have been major improvements over the last year; yet, as the platform matures, the expectations of the market change as well. SAP’s challenge is to couple the innovative character of the platform with the new responsibilities associated with cloud-based service delivery.

Note: Recently, there has been a decision to move the Hana Cloud downtime announcements to a mailing list. Although the reasons behind this move are primarily technical in nature,  I would like to see a more prominent publication of this material.  The mailing list looks like it is open to everyone but most other cloud vendors have a single page for planned maintenance (for example, here is the one for SalesForce) or the status of the platform (here is the one for SalesForce). I’d expect something similar for the HANA Cloud.

To report this post you need to login first.

3 Comments

You must be Logged on to comment or reply to a post.

  1. Harald Mueller

    Dick, not sure whether you are Dr. Watson or Mister Holmes. 😎

    As always you hit the bull’s eye. We are really sorry for the irritation.

    Internally we did discuss the upgrade of HANA to rev. 51 and the potential individual impact for the affected consumers. You can be assured that service delivery is key to our platform.

    I’m very much aware of the single pages of truth offered by other vendors. Knowing all the internals and details of our platform we are very much aware of the impact we have with our planned actions causing downtimes for each individual customer. We work on making his information and data very transparent and tailored to each individual customer. While one single – aggregated – page does create a certain outside perception it is not helpful for individual customer specific issues with the service.

    Harald

    (0) 
    1. Richard Hirsch Post author

      You can be assured that service delivery is key to our platform.

      I think this focus is critical for the platform’s success.  However, you guys are doing so much at so many different levels that I don’t know whether this topic is getting all the resources / attention that it really needs. Or you could be doing a lot in this area and none of this activity is being recognized / seen externally. The focus on technology (where the platform and team is strong) should be balanced with a focus on service delivery (where there is little information).  The public – especially in the enterprise –  needs to know more about your activities in this area.

      I’m very much aware of the single pages of truth offered by other vendors. Knowing all the internals and details of our platform we are very much aware of the impact we have with our planned actions causing downtimes for each individual customer. We work on making his information and data very transparent and tailored to each individual customer. While one single – aggregated – page does create a certain outside perception it is not helpful for individual customer specific issues with the service.

                         

      I like the strong customer orientation but I’m wondering how this model will scale as more customers get on the platform.

      I also had an interesting conversation on Twitter about the topic which you may find interesting. 

      D.

      (0) 
      1. Harald Mueller

        Well, you are right. That model clearly does not scale and the HANA update went through this exceptional internal process.

        We created some confusion with the announcement which was partly wrong.

        The Document Service has been updated but there was no impact on the consumers. It is running in a HA setup and the update was purely internal. This might give you some hints on the direction we are heading with the complete platform. Clearly a lot of work in front of us. Thanks for the hint on the twitter conversation, very interesting…

        Harald

        (0) 

Leave a Reply