by Kurt Schmidt, former head of the Information Technology Department (DFD-IT)
With the text that follows I am responding to the friendly request to supply a few remarks on the occasion of the 40th anniversary of DFD – making it by all means personal, if possible with a touch of humour, and perhaps focusing on the genesis of DIMS, in whatever way I see fit. DLR has been in my thoughts anyway, ever since I ceased working there; I drive past it almost daily. My memories are quite alive. For me, DLR is one of the felicitous institutions. I not only worked there, it was a big part of my life. And it is a pleasure to cast my eyes over the gently illuminated premises whenever I drive past in the evening: you might even call it the devotion of an engineer.
An anniversary is a time for an institution’s self-ascertainment by reflecting about its achievements, its institutional culture, and its robustness. It is also a moment of pride that I, too, feel strongly, because I was officially involved in projects for 34 years, plus additional years as ‘Bureau Qualitas GmbH’ building up a quality management system, and we have achieved a great deal.
My contribution will concentrate mainly on DIMS, the infrastructure that is at the heart of today’s Earth Observation Center. For the few who might not know: DIMS stands for ‘Data and Information Management System’. It is a universal data technology tool that can be used to process, archive and distribute in a standardised form the data flow from a wide range of remote sensing missions. There was a long road to its realisation. If I walk down that road again in my memories, please realize that my now considerable temporal distance from more recent events may somewhat distort reality.
The founding father of DIMS was Winfried Markwitz. He defined the aim in 1993 and steered developments in the desired direction over the first three difficult years until his retirement from DLR. As one of his department heads, I had his trust and support in its realization.
After a two-year interregnum under Wolfgang Mett, there followed with Stefan Dech and Richard Bamler the era of young leaders. The idea of a cluster was born. Things were seen in a larger context: becoming something more than an institute, the fascination of a higher level of integration, a new quality. DIMS was also part of this. And once again, now as head of the Information Technology department, I enjoyed their trust and support during its refinement and completion.
The definition of the aim was opportune. After the 1993 evaluation of DFD the tasks to be accomplished were clear. Mission-specific infrastructure was already successfully operating, but that was the Old World. The new challenge was migration into the New World. So DIMS turned out to be the right topic at the right time, and therefore a topic with a prospect of success.
However, even with the overall backing of the bosses, a number of issues crop up. Do we know enough? Do we have enough staying power? Are we ready for the adversities that life brings?
It’s a wonderful thing to enjoy the trust of the leadership. But trust is coupled with expectations. It then becomes important to keep the right distance to the core of the expectations in order to avoid being pulled into something like a black hole. Development requires having enough freedom for creative exploration. It’s a question of modelling something innovative: describing things as simply as possible, devising structures and sequences, imagining processes extending beyond the borders of departments and institutes, and above all, spending enough time thinking before any implementation begins. But now, each of these in turn.
At the German Space Operations Center (GSOC) and in the Applied Data Technology department (WT-DA) we had acquired the necessary maturity for the new, large task thanks to prior development commissions on behalf of DLR institutes, Max Plank institutes, and ESA.
The name of the department changed over the course of time from Systems Technology (GSOC-ST) to Processor Technology (WT-DA-PK) to Systems Development (DFD-SE) to Information Technology (DFD-IT), whereby the core team remained basically the same the whole time.
The tasks we faced before turning to DIMS were diverse. The most important ones are listed below as an indication:
The core team was accordingly experienced in dealing with scientists, had learned to be useful in a wide variety of external projects, and also to keep engaged in discussions and pugnacious debates, even when meetings extended into the late evening hours and were often not conducted in their native language. This made us actually well equipped for the new DIMS challenge, and we could have started right in, except for there being a precursor system that had the support of many colleagues.
The precursor was an archive with manually transported storage media, a Data Management System (DMS) in the form of a database with partially automated catalogue updating, together with an Intelligent Satellite Information System (ISIS) as a user interface. The cascading of processors to form processing chains was controlled by administrators.
This contrasted with the DIMS idea, which was to assume the form shown in the DIMS functional diagram that anticipates its later realisation. The heart is a robot-supported archive and the associated catalogue, a so-called Product Library. There is no manual transport of storage media, and consistency between archive and catalogue is assured automatically. Grouped around this core are the functional units data entry (Ingestion Processor), ordering (User Information Services coordinated with Order Management), and delivery (Online/Offline Product Generation & Delivery). The formation of processing chains is automatic via a production control function unit (Production Control) that initiates different processing systems (Processing System 1, 2, etc.). Satellite data are not directly archived, but only after the addition of metadata and conversion into a standard Level 0 format, by the data entry function (Ingestion System). Higher processing levels, Level 1 and Level 2, are always archived, but Level 3 output from the post-processing functional unit (Post-Processing) only on request. (The Earth elevation model derived from SAR data is an example of an archived Level 3 product.) The whole system is run using a convenient operator interface (Operating Tool).
DIMS functional diagram
In 1993 Mr Markwitz instructed the department to develop a new system to store, manage, offer, and deliver remote sensing data as a replacement for the previous system. ISIS was to be integrated as the user interface. As to DIMS, there was not yet a detailed model, far less any specifications, but the idea was nevertheless worth discussing to the extent that one could expect broad consensus. But that was not to be the case. DIMS initially met with little enthusiasm on the part of many department heads and colleagues. The struggle to master the technology, which characterized the preliminary phase of DIMS, now turned into a struggle against an insistence on remaining with the existing system. For a long time it was impossible to set up a proper project team. Many colleagues were to be involved from several departments, also including some from Neustrelitz, all of whom had limited time available for the purpose.
Thus it is no surprise that there was never a formally signed project plan for DIMS. Version 0.1 never got to 1.0, but instead remained a relatively flexible and open management tool. The intermittent involvement of the Space Management division in Cologne did not alter anything, nor did the involvement of Ulrich Walter (Planning and Large-Scale Projects division) as project head, along with a considerable expansion of the project and the new name, ‘Operational Data Use’. DIMS development was now only one work package among five. A year and a half later the project was again slimmed down. Only later, after the reorganisation of all the staff working on DIMS and a few associated tasks into a new department, DFD-IT, in 1998 did the project gain somewhat more consistency and tranquillity. As far as the project plan was concerned, on reflection it turned out to be consoling that a ‘live’ project plan kept under discussion but without official approval can lead to the goal more certainly than can a ‘quietly deceased’ Project Plan 1.0 buried in some file.
The reader should not get the impression here that a lax attitude about project plans emerged. That I certainly oppose: In the case of external projects the project plan is holy!
Another state of affairs was more serious than dealing with the project plan. When preparations got behind schedule at the ERS ground segment for processing data from the GOME (Global Ozone Monitoring Experiment) sensor, some of the core team was transferred from DIMS for as many as three years to work on GOME. External contracts had priority, which is still the case today.
These circumstances, along with the complexity of the task, were the reasons for the delay in developing DIMS. Everyone’s patience was put to a hard test. DIMS had to fight. A colleague concisely described this struggle as ‘winning wars of attrition’. Only in 1998, five years after its launch, did DIMS end up with more-or-less stable operation. From that point on, numerous satellite missions and sensors were connected to DIMS. The first mission completely managed with DIMS was SRTM (Shuttle Radar Topography Mission) in 2000.
At this point I can’t resist indulging in a light-hearted comment. It’s about DLR leadership. DLR does not really have, I think, a true power centre; instead, it has many of them, or better still, vegetation centres. Its culture is lifelike. Whatever grows is allowed to grow. Cultivation is done with a watering can. That also applies to the development of DIMS. There were two head gardeners, first Winfried Markwitz, then Stefan Dech. It worked because there was also weeding going on, and beds were reorganized, and because we knew that there was a rake in the shed. Also important was the fact that the gardeners were patient.
In the end this patience paid off. The DIMS team could manage to convince anew the leadership and critical colleagues at retreats and audits that the goal was no longer distant and the team in good shape. The three-day DFD leadership retreat in autumn 1999 in Frauenchiemsee had an agenda exclusively dedicated to DIMS. We remember it well.
The DIMS core team, in good shape and with DIMS documentation, 1st half-year of 1999
This photo was taken after a successful software review and shows, from left to right, Willi Wildegger, Eberhard Mikusch, Meinhard Wolfmüller, Markus Göhmann, Stephan Kiemle, Ralf Reissig, and humble me, seated. For this occasion, one of the many files with DIMS documentation is merrily featured. After all, according to the general maxim everything has to be ‘stringently recorded’. We came quite close to meeting that goal using UML modelling language and several crucial ESA engineering standards. The project required a System/Software Quality Assurance Plan (SQAP) and a Software Verification and Validation Plan (SVVP). Likewise an essential element of quality assurance was the separation of development and operating environments.
Any mention of names is always accompanied by the risk of not being fair to everyone and everything. I ask for understanding regarding the additional names I mention as I try to walk the line between historical comprehensiveness and light reading.
In the period up to my retirement from DLR in 2003 a number of additional important colleagues came to join us, like Christoph Reck, who was already involved in ISIS, and Peter Seifert, who contributed a great deal to operations, as did two ladies, Conny Bilinski and Daniele Dietrich.
When I think of Daniele, EOWEB always comes to mind (see ‘User Information Services’ in the DIMS diagram), a software package that she lovingly nurtured as if it were her own dear child, the perfect mother as far as EOWEB was concerned! As I reflect on each and every former colleague, I realize that that was actually a widespread attitude about their work.
Looking at DIMS today, one sees a firmly established bank, a property of historical dimension, a cultural heritage, indispensable for the work of the institute alliance. Stefan Dech put it this way at my retirement farewell gathering: “A sensor that isn’t connected to DIMS just doesn’t exist!” Could there be any better praise?
Before I conclude this account I’d like to share something close to my heart. The writer gets too much of the limelight, whereas DIMS is entirely the work of the clever people who created it. When I was writing this report, I contacted my former colleague Willi Wildegger, the good, reliable spirit of the house, to refresh my own recollections by tapping his archival memory. He didn’t agree with my protest that Edison’s conviction, that genius is one percent inspiration and 99 percent perspiration, was indeed true also for DIMS. After all, modelling at the top level, he said, continues down to the lower software levels, each step demanding a certain amount of genius, followed in turn by endless testing in the development world, followed by no less testing in the operational world. Quality is merciless. For example, the DIMS equivalent of Edison’s glowing, spiral-wound filament was, after all, a software effort of ten man-years just for the Product Library functional unit.
With this deeper insight into the matter, I connect my due gratitude for all those who contributed significantly to the genesis of DIMS.
The genesis can be considered basically concluded with the first mission entirely handled by DIMS, the SRTM mission of the year 2000, to the extent that any in-house project of this vitality can ever be considered completed. After all, the success story went on. Eberhard Mikusch assumed the leadership of DFD-IT in 2003, and as spiritus rector of the software, moved DIMS further along, together with the new project head, Stephan Kiemle.
To complete the picture, let me add that starting in July 2000 a DIMS milieu was also established at the National Ground Segment in Neustrelitz, with links to the receiving facilities and processing systems there, as well as a fast data link to Oberpfaffenhofen. Willi Wildegger’s counterpart in Neustrelitz was Hans-Jürgen Wolf.
What followed was a huge step as far as the range of DIMS was concerned. On June 7, 2004 a DIMS cooperation agreement was signed by DLR-DFD and Werum Software & Systems AG. This cooperation based on a joint code led to DIMS becoming a software product marketed by Werum. The main customer was ESA for their multi-mission facility infrastructure. Licences also went to Infoterra Deutschland (now Airbus Defence and Space), CSIR (Council for Scientific and Industrial Research) South Africa, and others. DIMS software was suddenly playing in another league.
This partnership with Werum is still alive and well, and DIMS is routinely adapted and migrated to new technologies.
Since 2007/2010 DIMS has been the data management basis for two extremely successful national earth observation missions, TerraSAR-X and TanDEM-X.
The Sentinel-5 Precursor satellite mission, launched in 2017, also runs entirely with DIMS at its ground segment.
And DIMS continues to write history beyond the short term: the hyperspectral satellite mission EnMAP (Environmental Mapping Analysis Program), with a launch planned for 2021, will entirely depend on DIMS for data management at the ground segment.
One can see that DIMS has remained astonishingly vital for a gradually aging system.