nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Performance: Opening FTA diagram

medini_1036016653

Dear support, a fellow user is struggling with medini's perfomance in a yet not communicated way: Double-clicking a FTA diagram takes several seconds, until it is loaded. Side-question: Is the license status checked during some operations, or only during startup and then cyclically after fixed time-intervals? Our network connection is currently not ideal. Best regards Peter

Find more posts tagged with

medini

medini legacy

Accepted answers

medini_1028038066

The opening of any diagram for the first time takes longer than any subsequent open. This is because of internal initialisation of the editors. For the issue described above there is a ticket and a feature request entered to the data base. Since there are more urgent performance issues, the discribed behavior will be investigated in future. Please make a not, if there is a higher priorty from the ZF audience.

All comments

medini_1024120963

Hi Peter, can your fellow please make a test and count the seconds until it actually shows up. For big FTAs this may take a little longer. It would be good if he can report this using the contact support button in the tool. My suggestion would be to initially load it and so sth. else until it is loaded. After the initial load everything, should come from the cache and therefore load much faster. BR Joël

medini_1036016653

Dear Joel, I sent the contact support inquiry: -- Dear support, as
requested in https://medini.freshdesk.com/support/discussions/topics/1000094546
I supply a log from a test I conducted. - I
freshly started v2019R1.1 - I
opened a FTA diagram using "return" key and at the same time took a
stop watch and measured 6 (six!) seconds for it to open - I
closed that FTA diagram - I
opened the same FTA diagram using "return" key and at the same time
took a stop watch and measured 1.8s for it to open Expected:
Opening a diagram should take no more than 100ms (one tenth of a second). The
diagram is not large. If requested, I add a screenshot
("SAFETY_L2_HW_LOGIC_6SO_3PS"). Thank
you for your help Best
regards Peter -- After all, a screenshot might be of interest, so here you go:

medini_1052680584

Cheers, I also cannot understand
why it takes that long: maybe medini builds a list of trees refering to the
diagram in background? I noticed the following: if I
have to add a transfer gate near the bottom of our big tree, it takes ages, because
medini just offers me the options which do not have logical loop (and for that an algorithm is necessary to run as I type). What I do to have a faster list of transfer gates offered to me: I first just
insert the transfer gate in the diagram without connecting it to another gate,
such that medini won't try to figure out the circular loop. Afterwards when I connect
it to the intended gate, medini tests again anyway if I am allowed to do it (and apparently much faster..). Regards, Andre

medini_1036016653

As replied by email: I'll gladly test the perfomance penalty that the network based license lookup might have on performance. Please provide a local (test) license, if possible.

medini_1036016653

Thank you for supplying a local test license. Performance regarding opening FTA diagrams is unchanged with that license type (5s initial opening, 2s for subsequent opening).

medini_1024120963

Hi Peter, we will release a new version of medini at Monday 06.04.2020. Especially the blocking issue with the slow calculation of a fault tree should be solved with this one. The opening topic is from my perspective more a minor topic as it is not enough time to get a coffee. It might be odd but after initial opening to wait for 2s until the diagramm shows up is for the majority of users still acceptable. One explanation why medini needs more ms than other solutions may be, that there is a big model behind which stands in relation to the events in the diagram. Other solutions you may compare it with do not have this model based approach. In general the medini team will try to improve the performance within the next years. Best regards Joël

medini_1028038066

medini_1036016653

Dear Krasimira, that is fine, at least from my point of view. Best regards Peter

medini_1024120963

Dear Peter, through investigations in your fault trees we have found scripts inside the events to provide the probability for instance. Did you once try to make a FTA tree without these scripts but with the same amount of elements? The current implementation is running these probabaility scripts everytime you open the tree. I wonder if you have even more scripts and if we could implement the behavior of these scripts in the long run. Do you have an overview of the scripts you use in regard to the FTA? The truth is scripts are great to make a customizations but they can cause also issues in regard to performance. Best regards Joël

medini_1036016653

Dear Joel, that script is not present anymore, in the last PHEV project I provided to Nino on last thursday. This used to be the only script that got executed during FTA diagram rendering. I plan to bring it back again, but wanted to rule it out as causing performance issues. In fact I do not see a difference in the speed of opening diagrams, as currently no diagram is showing any of the elements that used to contain a scripted probability. (This is another issue I reported regarding medini's inability to restore subtrees of helper trees PET:CE_U, AET:CE_U and SFT:CE_U.) As presented in the past, the script does a lookup of the related failure mode's diagnostic coverage in the DC worksheet. Best regards Peter

medini_1036016653

Dear Krasimira, personally, I still feel navigating through FTAs spanning multiple FTA diagrams very sluggish. I've heard similar feedback from internal colleagues and external partners. How is your current evaluation on priority of this "feature" request?

medini_1036016653

Dear support, was it in the meanwhile considered to improve sluggish UI performance, show-cased by opening/closing of diagrams? I just tested v2021 R2 beta on a High Performance computer and it feels the same. Thank you and best regards Peter

medini_1024120963

Hi Peter, usually you have very big FTA Diagramms. One major point that may practically support the opening time is the number of elements that need to be displayed at the diagramm. Did you already try with the latest version 22r1 if the issue is still there? If the issue is there for small trees? If the issue is there for big trees? What is time it needs for initial loading ? What is the time is needs after it was loaded initially? How many FTA events are we talking about? I am asking these questions to investigate if the issues is still there. BR Joël

medini_1036016653

Hi Joël, typically our FTA *models* are big, as they contain many objects diagrams / sub-diagrams. Each diagram on its own though rarely contains more than 50 elements. The slowness is also observed for diagrams with ~10 objects. Initial loading of diagram "#hWFTA#SNR_RPS01_FL" just took ~9 seconds (nine seconds). It comprises merely of an event (that is referenced by one transfer gate located on a different FTA model) and below a transfer gate to some sub-tree: Subsequent loading of the same diagram takes 1-2 seconds. After that I tried loading other diagrams for the first time. They take 1-4 seconds. Even when the diagram is already displayed, the mouse cursor keeps busy for another second. Closing also is taking several seconds. While the diagram itself disappears readily, it takes some seconds until the taskbar is refreshed. Expectation: On our modern hardware all of these operations should be felt as "instant" (<1s; may be 0.5s). My tests were conducted with PHEV project on v2020 R2. I don't have v2022R1 available to me. I'm rather positive that I had tested this also on v2021R2.1 and did not observe any optimization. You should have a project of ours available on your end to do some tests. If not I can make it again availabl to you (NDA applies). Best regards Peter

medini_1024120963

Hi Peter, thank you for your test. medini builts a failure tree model that it connects with. This may explain the thinking time of the tool. I hope that we may find sth to speed up the thinking time of medini. In order to have comparable conditions it would be beneficial to have the same project for reference. I will share a secure mail that you may use to share. NDA applies ;-)

medini_1036016653

Hi Joel, danke für Deine Rückmeldung in Bezug auf das
freshdesk-Thema Performance: Opening FTA diagram. Ich habe
das Projekt, das ich gestern getestet habe hier hochgeladen: Krasimira hat über Ihren Account ZF FileCloud Zugriff. Du
solltest über diesen Link drankommen: https://file-cloud.zf.com/index.php/s/iz9uLwN1tug2i76 PW for
encrypted ZIP: ANSYS_ZF PW for public
link: ANSYS_ZF0000! Die getesteten Diagramme waren unter anderem die im
Screenshot hervorgehobenen, ihr könnt aber auch die benachbarten oder beliebig
andere probieren. Schreib mir bitte kurz, ob das Beziehen der Datei klappt.
Der öffentliche Link ist für 14 Tage gültig. Krasimira kann über Ihren Account
dauerhaft zugreifen. ciao Peter